@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix dc: . @prefix skos: . vivo:departmentOrSchool "Applied Science, Faculty of"@en, "Mechanical Engineering, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Dong, Chunli"@en ; dcterms:issued "2009-02-24T00:00:00"@en, "1994"@en ; vivo:relatedDegree "Master of Applied Science - MASc"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description """Spray marking of fish, using fluorescent pigment propelled by compressed air, is a mass marking technique, and is useful in the studies of fishery management. This method typically uses a conveyor system to transport fish into a marking unit that dispenses the powdered pigment through a nozzle with the aid of compressed air. The marked juvenile fish are then released to continue their life cycle, and may be subsequently caught for the purpose of experimental investigation. The previous design of the marking machine (MK-5) had many shortcomings. In particular, the pigment dust caused worker discomfort and possible environmental pollution, overlapping of fish on the conveyor resulted in poor marking quality, and improper feeding and conveying caused injury and mortality in fish during marking. Due to these, the marking machine MK-5 was not widely used. In the present work, a new prototype of the automated spray marking machine was designed and built in the Industrial Automation Laboratory at the University of British Columbia. The new design was based on studies related to pigment emulsion, compressed air pressure, zig-zag shaped feeder channel, conveyor belt compartmentalization and pigment recycling. Image processing for assessing the retention of fluorescent pigment mark on fish body was investigated by first developing retention indices, and subsequently using appropriate image processing techniques such as the dilation and erosion of mark density, in order to recover an aged mark. Our design developments have led to a fast, accurate, reliable, low-cost and environmentally friendly system for marking fish. The prototype has been tested at several hatchery sites in British Columbia and has thus far exhibited good performance."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/4971?expand=metadata"@en ; dcterms:extent "3323618 bytes"@en ; dc:format "application/pdf"@en ; skos:note "D E V E L O P M E N T O F A N A U T O M A T E D M A R K I N G M A C H I N E F O R L I V E F I S H W I T H A P P L I C A T I O N I N F I S H E R Y M A N A G E M E N T By Chunli Dong B.A.Sc, Northwestern Polytechnical University, Xi'an, China, 1983 A THESIS SUBMITTED IN PARTIAL FULFILLMENT O F THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF A P P L I E D S C I E N C E in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF MECHANICAL ENGINEERING We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA January 1994 © Chunli Dong, 1994 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Mechanical Engineering The University of British Columbia 2324 Main Mall Vancouver, Canada V6T 1Z4 Date: f^b. 7 / / ^ f 4 A b s t r a c t Spray marking of fish, using fluorescent pigment propelled by compressed air, is a mass marking technique, and is useful in the studies of fishery management. This method typically uses a conveyor system to transport fish into a marking unit that dispenses the powdered pigment through a nozzle with the aid of compressed air. The marked juvenile fish are then released to continue their life cycle, and may be subsequently caught for the purpose of experimental investigation. The previous design of the marking machine (MK-5) had many shortcomings. In particular, the pigment dust caused worker discom-fort and possible environmental pollution, overlapping of fish on the conveyor resulted in poor marking quality, and improper feeding and conveying caused injury and mortality in fish during marking. Due to these, the marking machine MK-5 was not widely used. In the present work, a new prototype of the automated spray marking machine was designed and built in the Industrial Automation Laboratory at the University of British Columbia. The new design was based on studies related to pigment emulsion, compressed air pres-sure, zig-zag shaped feeder channel, conveyor belt compartmentalization and pigment recycling. Image processing for assessing the retention of fluorescent pigment mark on fish body was investigated by first developing retention indices, and subsequently using appropriate image processing techniques such as the dilation and erosion of mark density, in order to recover an aged mark. Our design developments have led to a fast, accurate, reliable, low-cost and environmentally friendly system for marking fish. The prototype has been tested at several hatchery sites in British Columbia and has thus far exhibited good performance. Table of Contents Abstract ii Table of Contents iii List of Tables v List of Figures vi List of Symbols viii Acknowledgement xii 1 Introduction 1 1.1 Background 1 2 Design Deve lopment of the Spray Marking Machine 6 2.1 Introduction 6 2.2 The Spray Gun System 6 2.2.1 Mark Retention 11 2.3 The Feeder and Conveyor System 15 2.3.1 The Conveyor 16 2.3.2 The Feeder 26 2.3.3 The Feeder and Conveyor System 37 2.4 The Pigment RecycHng System 40 m 3 Mark Retent ion Study 44 3.1 Introduction 44 3.2 System Hardware and Techniques 45 3.3 Basic Image Processing 46 3.4 Two Indices of Mark Retention . 49 3.5 Pigment Mark Recovery 52 3.5.1 The Square Dilation and Square Erosion Method 54 3.5.2 The Square Dilation and Plus Erosion Method 56 3.6 Discussion 59 4 Conclusions and Recommendat ions 61 4.1 Design of the Automated Spray Marking Machine 61 4.2 Field Tests 62 4.3 The System for Mark Retention Detection 63 4.4 Emulsion Concentration 64 4.5 Compressed Air 65 4.6 Spraying Aerodynamics 65 4.7 Marking of a Code 65 4.8 Pigment Mark Retention 66 4.9 Pigment Mark Detection 66 A Experimental Data on Zig-zag Shaped Feeder Channel 67 B S D S E M e t h o d of Connection between Pixels {k = 2) 77 C S D P E M e t h o d of Connection between Pixels (k = 2) 80 Bibl iography 85 IV List of Tables 2.1 Tests with Emulsion Spray 13 2.2 Test Data for Determining the Damping Coefficient of fish (unit: mm/s) 21 3.3 Values of the Mark Retention Index in the Experiment 59 A.4 Experimental Conditions 67 A.5 Dimensions of the Test Fish 68 List of Figures 1.1 The Spray Marking Machine MK-5 3 2.2 The Internal Structure of the Spray Gun 7 2.3 The Spray Cone 9 2.4 The Marking Areas of Spray Gun 10 2.5 The Conveyor 17 2.6 The Pile of the Simulated Dead Fish 18 2.7 The Model of Fish on a Sloped Conveyor under Viscous Damping . . . . 19 2.8 The Damped Motion of a Fish on a Smooth Plate 21 2.9 Speed-Slope Data for Designing the Sloped Conveyor Segment 25 2.10 The Feeder with a Zig-zag Channel 27 2.11 The Zig-zag Channel 30 2.12 The Uniform Distribution of Fish Orientation 31 2.13 Test Results on Fish Orientation Distribution 33 2.14 Experimental Results on the Dynamic Orientation of Fish 36 2.15 Experimentally Determined Dynamic Distribution Index pd{t) 36 2.16 Representative Distribution Index for a Design Prototype 39 2.17 A Schematic Diagram of the Pigment Recycling System 41 3.18 The Block Diagram of the System for Mark Retention Evaluation . . . . 46 3.19 The Histogram of the Image under Ambient Light 48 3.20 The Histogram of the Image via the Green Channel 49 3.21 The Histogram of the Image via the Red Channel 50 vi 3.22 The Parameter Block 53 3.23 The Three Symmetric Parameter Blocks 54 3.24 An Example of Binary Image Processing in Dilation and Erosion 54 3.25 An Example of Image Processing with SDSE 55 3.26 The Connecting Region of a Bright Pixel 57 3.27 An Example of Pigment Mark Recovery 58 VII List of Symbols ^1,2 ,3 The regions in a histogram diagram b The pixel brightness of an image -Bi ,2 ,3 ,4 The regions in histogram diagrams Ci The clearance of Stage 1 of the zig-zag shaped channel C2 The clearance of Stage 2 of the zig-zag shaped channel d The linear viscous damping coefficient En The pixel light energy of the n-th fish f{b) The pixel density with different brightness f{v) The damping force f{6) The distribution of fish orientation Fi The force along the inclined conveyor surface F2 The force normal to the inclined conveyor surface g The acceleration due to gravity Hj The average height of fish k The number of convolutions in image processing K The pixel quantity vui m The mass of an object Ml The location of the fish orientation measurement in Stage 1 Ma The location of the fish orientation measurement in Stage 2 n The order number of an individual fish Tic The distance along a column direction between two pixels Tir The distance along a row direction between two pixels N The total sampling number Pio . • . The supply air pressure to the spray gun, with reference to ambient conditions Rb The retention index in a batch of fish R, The retention index for an individual fish s The distance between stages 1 and 2 of the zig-zag shaped feeder channel S The distance from the tip of the spray cone to the target S' The distance from the spray gun nozzle to the target Sa The average fish body area in a batch Sf The fish body area Sn The body area of the n-th fish ^p The area of pigment particles on a fish body t Time Tf The average thickness of fish IX Uo The exit speed of the spray from the gun V The final static speed Vfc The fish sliding speed in the bottom layer Vc The maximum speed of the conveyor Vd The maximum delivery speed Vfn A speed constant Vt The fish sliding speed in the top layer a The angle of inclination of the plate with respect to ground am The maximum natural slope angle a , The angle of the sloped segment of the conveyor 6 The fish orientation angle 6m The diameter of the marking area on the target 0, The angle of the spray cone (p The angle between the feeder bottom wall and the ground ^c The diameter of the driving wheel of the conveyor (pm The diameter of the marking area of the target , The diameter of the spraying area of the target p The distribution index pi The distribution index at Stage 1 p2 The distribution index at Stage 2 Pd The distribution index from experimental data Pd{i) The dynamic distribution index from experiment Pf The distribution index on the conveyor belt with flanges T Time constant of dynamic distribution index on a plate Tc . . . Time constant of dynamic distribution index on conveyor with rubber cylinders a The standard deviation of fish orientation XI A c k n o w l e d g e m e n t I wish to thank Dr. C. W. de Silva and Dr. R. Cosine for their strict supervision and guidance as well as financial support throughout this research. This work has been supported by a grant from the B. C. Hydro and Power Authority, and was carried out in collaboration with the B. C. Ministry of Environment. Additional support has been provided by the Natural Sciences and Engineering Research Council. The assistance of Mr. Ken Ashley and Mr. Brian Ludwig of the B. C. Ministry of Environment, and Mr. Jimmy Yang of the Industrial Automation Laboratory is greatfully acknowledged. Also, I wish to thank Mr. Kanji Tsumura, Mr. Morley Rempel and Mr. Larry Mitchell of the Fisheries Research Facility at the Eraser Valley Trout Hatchery (FVTH) in B. C. for their generous and constant help in testing the prototype automated spray marking machine. Special thanks should go to Dr. N. Robin Liley, professor in the De-partment of Zoology, for providing a fish tank and for his technical advice in maintaining live fish. xn This is dedicated to my parents Fengyun Song and Huali Dong xni Chapter 1 Introduction The objective of the present work is to design and develop an automated, fluorescent-pigment spray marking machine for fish. This is based on an established fish mass-marking technique, and many problems have been experienced when using the technique in the past. Fish marking is known to be useful in the studies of fishery management. Our work will address various fish marking techniques and will design, construct and test a prototype fish marking machine for use in field operation. 1.1 Background Fish marking is a very important and useful method in studying the population dynamics, migration patterns, dispersal and other aspects of the life history of fish. With the advent of modern fishery, it has become more important to use fast, convenient, reliable and low-cost marking techniques in fishery management. This is particularly true in view of dwindling fish stock that we see in Canada, particularly as a result of overfishing, primarily due to poor techniques for predicting fishery stocks. There has been significant research on fish marking since 1950. Many popularly-used fish marking techniques include coded wire tags (CWT), clipping, freeze brands, chemical brands, electronic tags, and fluorescent pigment application. For example, the CWT method injects a coded wire tag into a bony organ, typically the nose region of the fish. When doing this, the operator must hold a fish tightly by hand, position it 1 Chapter 1. Introduction 2 properly with respect to the injector, and inject the tag using a mechanical device. The entire process can be slow. After releasing the tagged fish into water, some tags can be dislodged and lost. A main advantage, however, is that a wire tag is able to carry the required amount of information in a coded form. Similarly, the other methods also have advantages and disadvantages. With respect to marking speed, cost and the survival ratio of the marked fish, the fluorescent pigment marking technique is considered superior. The fluorescent pigment mass marking was first studied by Jackson in 1959 [1]. It was proven successful with salmonids (Phinney, et al., 1967 [2]; Hennich and Tyler, 1970 [3]; Phinney, 1974 [4]) and warm water fish (Ware, 1969 [5]; Pierson and bayne, 1983 [6]). Phinney et al. [2] reported that this technique was fast and inexpensive, the fish survival was excellent, and that a mark retention was 130 days or more was possible. After air powered devices for dispensing fluorescent marking pigment came into use, the marking speed has been improved significantly (Reports indicate that marking rates of more than 100,000 fish per hour are possible). The spray marking technique uses high pressure air to embed microscopic fluorescent granules into the epidermis of fish. The granules may retain for several years, and may be observed by examining the fish under ultraviolet light ( Moodie and Salfert, 1982 [7]; McAfee and Loucks, 1986 [8]; Pauley and Troutt, 1988 [9]). This method typically uses a conveyor system to transport fish into a spray marker that dispenses the powdered pigment through a nozzle with the aid of compressed air. An agitator is used to shake the container of dry pigment in order to reduce clogging of the nozzle. The marks are detected in the field by examining samples of fish under a low-power ultraviolet (UV) lamp. During examination under a UV light source, the spray-marked areas on the fish body, that contain the fluorescent particles, will shine brightly in a specific color such as red, green, or orange in the visible spectrum and these marks can be very easily detected through the naked eye or by optical means. Chapter 1. Introduction Spray Gun Dye (Dry Powder) Compressed Air (Hight Pressure) //////////////////97////////////////M V///////// Figure 1.1: The Spray Marking Machine MK-5 The ability to mark large quantities of fish at high speed, low cost, and with high reliability is quite beneficial in the evaluation studies of fishery programs. For instance, manual marking by hand clipping can cost about $0.05 per fish, with a typical capability of 6000 fish per day. Hence, marking of a batch of 400,000 fish will cost about $20,000 and will require a team of three people working for 22 days. It is estimated that , with machine marking, a similar batch of fish may be marked in about 10 hours by the same number of people. The associated speed increase is about twenty fold and the cost reduction is about as large as 94%. The machine model MK-5 is an automated pigment spray marking machine for fish. A schematic diagram of this system is shown in Figure 1.1. This machine was developed Chapter 1. Introduction 4 recently in Fishery Research Institute of the University of Washington [10]. Assuming that there is no delay in feeding fish into the machine, about 150,000 fish/hr can be marked using approximately one pound (0.4536kg) of fluorescent pigment per 7,000 fish. The size of the fish ranges from 38 mm to 52 m,m. in length. Due to the dry powder of the pigment employed, many drawbacks exist and these can seriously degrade the performance of the marking machine. During the operation of the machine, a significant amount of pigment escapes into the air. This wastes the pigment material and creates an uncomfortable working environment for the operators. Furthermore, because the conveyor belt is flat and fish are dropped from the feeder directly onto the conveyor, an overlapping problem can occur, thereby resulting in poor marking quality. When marked fish are released into water, they may carry an excessive amount of pigment which simply sticks onto the fish body. This can result in contamination of fish tanks, ponds and the environment around us, by the marking pigment. This environmental issue has not been thoroughly studied, but, nevertheless it is desirable to reduce the quantity of pigment that enters the natural water system. Juvenile fish that are hatched in hatcheries will complement the natural fish stocks. All the hatchery fish are expected to be marked before they are released into the habitat. The fish will be harvested in two or three years at various locations. Among the harvested samples of fish, the ratio of the artificially hatched fish to the natural born fish can be found by identifying the pigment mark on fish. The aim of the machine that is developed in our work would be to overcome and eliminate these drawbacks of previous methods of marking and to develop a fast, accurate, reliable, low-cost and environmentally friendly system for marking fish. An improved design for an automated marking machine for fish is described in Chapter 2. Key parts of the machine, such as the spray gun system, the feeder and conveyor system, and the pigment recycling system are described. The design of a mark retention Chapter 1. Introduction 5 observation system to evaluate the mark quality of fish is described in Chapter 3. The new marking machine is also further studied on Chapter 3, by considering image processing techniques, such as light arrangement, thresholding on image analysis, fish mark indices, and the pigment mark recovery. Conclusions and recommendations are presented in Chapter 4. Chapter 2 Design Development of the Spray Marking Machine 2.1 Introduction The spray marking machine MK-5, as schematically shown in Figure 1.1, has many drawbacks as noted in Chapter 1. An improved design of the machine must provide a fast, accurate, reliable, low-cost and environmentally friendly system. In order to overcome problems such as overlapping, pigment dust, and pigment wastage, the system is designed to consist of three subsystems: the spray gun system, the feeder and conveyor system and the pigment recycling system. Briefly, the research here will focus on the improvements to the MK-5 machine and on enhancing the features of each subsystem of the existing machine. 2.2 The Spray Gun Sys tem The spray gun system includes three parts: 1). The spray gun; 2). the pigment container; 3). the compressed air cylinder or compressor. The internal structure of the spray gun is schematically shown in Figure 2.2. When the compressed air goes through the gun, powdered or liquid pigment from the container is drawn into the gun because of the vacuum effect. The pigment exits the nozzle as a Chapter 2. Design Development of the Spray Marking Machine I Compressed Air Pigment Vacuum Effect Sprayed Pigment Figure 2.2: The Internal Structure of the Spray Gun spray, at a very high speed. The fluorescent pigment is a grit-based dye, with a particle diameter ranging from 50 to 350 fim. The pigment is known to be nontoxic, and it is visible under ultraviolet light. It is composed of transparent organic resin particles which are capable of fluorescing in solid solution. A pigment emulsion has been substituted for the dry powdered pigment which is used in the MK-5 machine, in the new design of the machine. Unlike the case of dry powder, much of the sprayed emulsion is collected in a container attached to the underside of the Chapter 2. Design Development of the Spray Marking Machine 8 machine at the exit, and will be recycled, providing a degree of environmental friendliness. Water in the emulsion increases the momentum of the pigment spray during marking, thereby increasing the mark retention. Also an emulsion facilitates the use of the system under damp conditions, thereby reducing nozzle clogging. The overall system is easier to operate than the original MK-5 machine. Although the particles of the fluorescent pigment spray impinge on fish body with a high momentum, only a fraction of the pigment particles actually embed into the scales of fish. From testing we know that the angle of the cone of the spray is about 16°. The distribution of the particle speed inside the cone is as shown in Figure 2.3. The spray region is the area of intersection of the spray cone and the target, as shown by the bottom section of Figure 2.3. The pigment speed is close to zero along the edge surface of the spray cone, in view of the static conditions of the ambient air of the spray jet. Furthermore, the concentration of the spray near this boundary layer is also low, due to mixing with the ambient air. Hence, since only the pigment particles above a certain speed and certain concentration embed into the scales of a fish, the marking region is actually smaller than the overall spray region. As shown in a dotted line, the marking region increases as the distance between the nozzle and the target increases, but the concentration of the spray decreases (by the principle of mass conservation). Hence, the marking region actually begins to decrease beyond a certain depth and will reach zero at a particular depth. An experiment was carried out with 120 psi air pressure applied through a piece of rubber hose to the spray gun. Spraying a target positioned 30 cm from the nozzle resulted in a 8.5 cm spray region diameter and a 5 cm marking region diameter. This experiment will be used as the basis for the spray arrangement for later studies. There are many factors which influence the marking region, such as the length of the hose from the compressed air cylinder to the spray gun, the nature of the valves Chapter 2. Design Development of the Spray Marking Machine Spray Gun Nozzle Figure 2.3: The Spray Cone Chapter 2. Design Development of the Spray Marking Machine 10 a) With a single gun system b) With a triple gun system Figure 2.4: The Marking Areas of Spray Gun that are located along the hose, the diameter of the orifice of the nozzle. The subtle relations among these that dominate in a complex manner, particularly in relation to jet aerodynamics, will not be studied here. The conveyor belt is roughly 15 cm wide. According to the typical experimental conditions, one spray gun is able to cover only a 5 cm width. A triple-gun system is needed in the new design, to produce the required marking area on the conveyor, as shown in Figure 2.4 (b). The MK-5 Machine has a single-gun spray system. We assumed that the ratio of the width of the marking region to the depth of the conveyor belt from the spray nozzle is to be constant at 5/30 in the above-mentioned experiment. For the single gun system, the depth has to be increased 3 times, to 90 cm,, in order to cover a 15 cm, width of the conveyor belt, as shown in Figure 2.4. To keep the same momentum of the pigment Chapter 2. Design Development of the Spray Marking Machine 11 particles at the conveyor belt, the coming out speed (UQ) of the pigment has to be increased 3 times, and the pressure (Ao) of the compressed air to the spray gun has to be increased 9 times according to the aerodynamic equation: Piocxul (2.1) In comparison with the triple gun system, the marking region produced with a single gun system is redundantly thick, and the compressed air and the fluorescent pigment consumed will be 3 times that of the triple gun system with little or no improvement in marking quality. From the discussion, we concluded that a single gun system is not as efficient as the multi-gun systems for the same marking requirement in the new design. Instead of the dry powdered pigment which is used in the MK-5 machine, a pigment emulsion has been studied in the new design. During spray marking, only a negligible quantity of the sprayed pigment emulsion escapes from the machine into the air, when an emulsion is used. This pigment drops onto the ground within a range of one meter from the machine. Most of the sprayed pigment emulsion sticks to the conveyor belt, then drops to the bottom of the machine, where it is collected in a container and recycled. 2.2.1 Mark Retent ion Early studies (in 1967 [2]) reported a mark retention duration of more than 130 days, and later studies (in 1982 [7]) reported longer retention durations of several years. These past studies are based on data collected using spraying tests with dry powdered pigment. The tests for mark retention done using the redesigned machine are described below, to evaluate spray marking with a pigment emulsion — an innovative approach to spray marking. Chapter 2. Design Development of the Spray Marking Machine 12 Four tests were completed at the Fisheries Research Facility at the Fraser Valley Trout Hatchery (FVTH) in July, 1992. The mark retention results of the tests are listed in Table 2.1. Spray pressures of 120psi and 175psi, and dye-water ratio of 1:1, 1:2 and 1:3 have been used in these tests. The mean length of the tested rainbow trout was 12cm in July, 1992 and these fish grew to 25cm by May, 1993, probably due to overfeeding! The sprayed pigment marks were checked manually under the ultraviolet light system and classified manually into one of the three categories: \"High\", \"Medium\" or \"Low\", depending on the retention level. An observation level of \"High\" retention means the pigment mark on the fish was distributed across the body and could be seen easily. For the observation level of \"Medium\", the pigment mark also could be seen easily, but in smaller regions or concentrated in spots. For the observation level of \"Low\", the pigment would not be clearly evident in the fish body. There are two things which have to be considered here: 1). The observations are quite different among checks at different times and can be subjective, when made by different persons. The subjectivity is largely due to the uncertainty of human visual perception and judgement. Without a quantitative measurement, the mark grades cannot be differentiated easily between the level \"High\" and \"Medium\". This problem has been solved in the later tests with the mark retention observation system described in Chapter 3; 2). Some fish in a particular batch had not been marked properly. Because the batch size was quite small and the marking speed was high, some fish had gone through the spray marking area on the conveyor belt, before pigment emulsion had time to reach the conveyor from the gun nozzle. These tests were carried out just for the mark retention studies. The rate of marking itself is not particularly important in these tests. According to the tests, the survival rate with 120 psi compressed air pressure is higher than that with 175 psi. On the other hand, spraying at 120 psi pressure would save much more compressed air and marking dye than at 175 psi. Because the difference in marking Chapter 2. Design Development of the Spray Marking Machine 13 Table 2.1: Tests with Emulsion Spray Batch Size 74 66 63 89 Dye:Water Ratio by Weight 1:1 1:1 1:3 1:2 Spray Pressure (psi) 120 175 175 175 Retention Duration (weeks) 1 2 16 31 43 1 2 16 31 43 1 15 30 42 1 15 30 42 Number of Fish in Each Mark Category High 67 58 46 45 41 54 51 33 33 41 42 20 34 28 55 34 37 38 Medium 5 12 17 18 19 8 10 12 12 9 10 15 8 12 14 19 14 23 Low 2 4 4 4 6 4 5 10 9 5 11 20 13 14 20 27 28 20 Survival No. 74 74 67 67 66 66 66 55 54 55 63 55 55 54 89 80 79 81 % 100 100 90.5 90.5 89 100 100 83 82 83 100 87 87 86 100 90 89 91 Chapter 2. Design Development of the Spray Marking Machine 14 quality is not significant in these tests, the savings associated with low pressure would be desirable and strongly recommended. The sprayed mark quality with dye-water ratio of 1:1 is the best comparing with that with dye-water ratio of 1:2 and 1:3. It also was found that the emulsion would be too dense to be sprayed if the dye-water ratio is higher than 1:1 (for example, 3:2). Finally, at a spray pressure of 120 psi, a dye-water mixture ratio of 1:1 for the pigment emulsion, and a 30 cm height the nozzle from the conveyor belt (the target), were chosen as the operating specifications for the new design. There are some questions which still remain to be answered through tests. For ex-ample, the effects of the size of the hose from the air pressure regulator to the spray gun, and the size of the nozzle orifice to the marking quality have not been extensively studied. Also, the aerodynamics of the spray system has not been systematically in-vestigated. Because the conveyor belt is within the spray cone, both the spray region and the marking region will change to some extent due to the associated fluid dynam-ics, even though the conveyor belt can move through surrounding air without significant aerodynamic resistance. A standard Milton Hydro-air Washer is chosen as the spray gun component for the design of the automated spray marking machine. The tests have been done with this spray gun to investigate the relationship between the spray pressure and the mark retention. A spray gun with multi-nozzle jet may be developed in the future. A more complete study in this area may consist of: Investigation of the relation between the spray pressure and the spray speed at the nozzle; investigation of the spray cone and the speed or momentum of pigment particles based on the speed profile within the spray cone; and the study of the rate of pigment particles, which are actually lodged in fish skin or scales to form the mark, based on particle speed and the structures of the fish skin. Such future work is recommended in Chapter 4. Chapter 2. Design Development of the Spray Marking Machine 15 For the future reference, the equipment used in the tests has the following specifica-tions: Hose: 9.5mm in diameter and 5 TTI in length; Spray Gun: (Milton Hydro-air Washer), 4.5mm diameter nozzle orifice. 2.3 T h e Feeder and Conveyor Sys tem The overlapping problem of fish on the conveyor belt during spray marking in the MK-5 machine degrades the marking quality. The primary parts of the existing machine which contribute to this problem are the feeder (or hopper) and the conveyor belt. These two parts are studied as a single system consisting of the feeder and the conveyor. To eliminate the overlapping problem, the conveyor has to be inclined with respect to the horizontal. The effect of gravity on the fish behavior in the new design of the conveyor is studied in this section. For the feeder, a zig-zag shaped channel, as shown in Figure 2.10, is employed. This will regulate the flow rate of fish as they exit the feeder and reach the conveyor, and will constrain the fish in an orientation particular to the direction of motion of the conveyor. This feature is expected to enhance the marking quality, and may be used for accurate coding of fish in the future, not simple binary marking. The present study is concluded with experiments on random fish motion. Chapter 2. Design Development of the Spray Marking Machine 16 2.3.1 The Conveyor The Construct ion The conveyor is used to carry fish from the feeder, through the spray marking region, to the exit of the marking machine. The conveyor is flat in the MK-5 machine, and the conveyor belt is a flat wire mesh, with no compartments. After dropping onto the conveyor, fish are free to move. If more fish come from the feeder, they will fall on each other when going through the marking region, thereby causing an overlapping problem. As a result, they could not be marked evenly. Furthermore, some fish on the conveyor are often pushed opposite to the direction of the conveyor motion, by the high speed air gun, which can result in jamming of the fish near the feeder exit. The marking quality, therefore, is difficult to control with the old version of the machine. In the new design, the conveyor is mounted as shown in Figure 2.5. The initial segment of the conveyor is sloped up. Fish are loaded to a position near the bottom of the conveyor, and they are carried by the conveyor along the slope, before transporting horizontally through the marking region. Flanges or a matrix of rubber cylinders have been installed on the mesh of the conveyor belt. The height of the flanges is such that only one layer of fish occupies the conveyor as it moves up the slope. Any fish on the top of this layer will slide down to either the bottom of the belt or into the vacant compartments on the belt. There is a stationary upper mesh of wire just above the conveyor compartments in the horizontal region of the conveyor. Just before being carried to the horizontal section of the conveyor, the fish pass under this metal mesh guard which restricts their movement. In this way, fish remain in one layer when moving through the marking region until they reach the exit of the machine. The fish are secure and will not be affected by the spray, as in the case of the old design. As discussed above, the sloped section of the conveyor is crucial for eliminating the Chapter 2. Design Development of the Spray Marking Machine 17 Feeder Marking Section Guard Conveyor Belt Flanges Figure 2.5: The Conveyor overlapping problem. The sloped section is examined by considering the gravity effect. T h e Gravity Effect As with any type of powder or granular material, a natural slope angle exists, under the gravitational field of Earth. The natural slope angle determines how much material can be put in a single pile, and how high that pile would be. The actual slope angle cannot be greater than the natural slope angle; otherwise, the material will slide down along the slope until the slope angle reaches the natural slope angle. The natural slope angle depends on the configuration and surface friction of each piece or particle of the material. The natural slope angle of fish has been investigated in the Industrial Automation Laboratory. The result is that the natural slope angle of fish is zero because of almost zero Coulomb friction on fish skin, besides the viscous-type damping due to the mucus. If we put a small number of unfrozen dead fish ( about 50 fish ) into a pile, the height of the pile will come down slowly until, theoretically, the slope angle becomes zero. The speed depends on how large the pile is. If we put a small group of live fish into a pile, the pile will collapse down in a very short time, due to random, jerky motions,of the Chapter 2. Design Development of the Spray Marking Machine 18 Figure 2.6: The Pile of the Simulated Dead Fish individual fish. From the analysis, a non-zero natural slope angle should exist according to the con-figuration of fish. Dead fish can be piled up as schematically shown in Figure 2.6. The slope angle a^ can be calculated, approximately, as am = arctan(—-) (2.2) where Tf = the average thickness of fish; Hf — the average height of fish. For rainbow trout, we empirically set T/ = 14 and Hj ~ 29 with a corresponding a ^ of 25.8°. Because the slope angle can never be greater than this limit, the slope angle value obtained in this manner is called the maximum natural slope angle. In the real situation, however, fish cannot be piled up as evenly or as high as illustrated in Figure 2.6. The practical natural slope angle in a dynamic view is much smaller than that found as above. The practical value is found according to the settling down behavior of the pile, as follow: placing a group of live fish (about 50) into a vertical tube and set it Chapter 2. Design Development of the Spray Marking Machine 19 -^-A Figure 2.7: The Model of Fish on a Sloped Conveyor under Viscous Damping upside down on a flat surface. When the tube is lifted, the pile of fish begins to collapse down. As expected, the speed increases quickly at the very beginning. When the slope angle reduces to an intermediate value, the speed begins to slow down because of the resistance of the mucus on the fish skin. Finally, fish on the top of the pile will slide down slowly until all the fish lie on the surface in a single layer. This intermediate slope angle is the practical natural slope angle, and is about 13.8°. The whole procedure lasts about 0.1 second. In addition to the natural slope angle, the damping friction caused by the mucus should be considered in the behavior discussed above. Assuming that the conveyor belt is flat, and without any flanges or compartments, the movement as fish slip down the inclined conveyor belt can be modeled as a block on a inclined plate, as schematically shown in Figure 2.7. Because only the damping friction exists between the contact surfaces, the force balance equation at the final static speed v can be described as Fi = fiv) (2.3) whe re Fi = the gravity force along the plate surface; / ( f ) =: the damping force. Chapter 2. Design Development of the Spray Marking Machine 20 Suppose that the viscous damping force is linear with the speed. The damping force can be expressed as f{v)^vd (2.4) where V = the final static speed; d = the viscous damping coefficient. From Figure 2.7, we have Fi = m g s i n a (2-5) where a = angle of inclination of the plate to the horizontal; m — mass of the object. Combining equations ( 2.3), ( 2.4 ) and ( 2.5), the final static speed can be expressed as V = —— sina = Vm sin a (2-6) a where Vm = speed constant given by ( ^ ) . Tests were carried out with 12 cm long fish (live rainbow trout) on a smooth plate at different angles to the ground. The data are given in Table 2.2 and are plotted in Figure 2.8. In carrying out the test, first, a number of dark lines were drawn on the inclined Chapter 2. Design Development of the Spray Marking Machine 21 Table 2.2: Test Data for Determining the Damping Coefficient of fish (unit: mm/a) a° ^(1) ^(2) V(3) ^(i) VIS) 10 398 570 470 600 20 1050 950 1000 1000 1050 30 1560 1530 1620 1600 1600 40 1900 1900 2000 2000 2200 50 2200 2500 2500 2300 2600 60 70 2500 3000 2500 3000 2400 3000 2700 3000 2600 3000 Speed 4 V (m/s) 3 -2 -1 -O • ^ 0 * % -s^\"^-T 10 20 30 40 50 ~i 1 1 r~^ 60 70 80 90 Slope of the plate (deg,) a Figure 2.8: The Damped Motion of a Fish on a Smooth Plate plate at one centimeter apart in parallel as reference positions of fish sliding down the plate. The experiments were videotaped and the line passing times of the fish were timed. From this sample of data, the constant Kn is established as approximately 3200 vfimjs. Further tests with real conveyor belts, instead of a smooth plate, have been done. The sliding speeds for fish in the bottom layer as well as for those on the top layer for 22° and 34° slopes are illustrated in Figure 2.9. The speeds for comparable angles of the conveyor are much smaller than that for the smooth plate. In the case of the conveyor belt, with either flanged compartments or rubber cyHnder grid, the practical speed constant (Kn) is found to be about 800 mm,js for the top layer, and less than 200 mmjs for the bottom Chapter 2. Design Development of the Spray Marking Machine 22 layer. Conveyor Speed The belt wheel close to the exit of the conveyor is the driving wheel, and is driven by a 1/2 HP DC motor through a worm gear. The driving wheel speed is adjustable from 0 rpm. to 45 rpm,. The diameter of the driving wheel (denoted by (f>c) is 92 mm,. The maximum speed of the conveyor belt is given by K - ^ ^ ^ ^ T T , ^ , = 217(mm/3) (2.7) oU The speed of the conveyor is adjustable from 0 mmjs to 217 m,m/s. In the flanged compartment configuration of the conveyor belt, the space between adjacent flanges (com-partment width) is about 35 m,m,. If each flange carries one fish, the maximum delivery speed is given by V, = ^^J (^\"^ /^ ) = 6.2(//5) = 22,320{f/h) (2.8) 35(mmj The marking speed of the MK-5 machine is reported as 150,000 fish/hr, with a 4.5 cm mean length (about 6 TTITTI of width) of chum salmon fry. The fish size is much smaller than what can be accommodated in each compartment in the new design. In other words, the compartments of the new design is suitable for the yearling fish with a 12 cm, length. So both marking speeds in the new design and for the MK-5 machine becomes comparable if concerning the size difference between fish and compartments. Considering the marking quality, a very high delivery speed may result in poor quality marking of fish. As observed in the last section, the maximum speed of the conveyor Vc (217mm/s) is only one quarter of the maximum damped speed of the top layer (Kn/4). So, the slope section of the conveyor may be designed easily within the speed limit. Chapter 2. Design Development of the Spray Marking Machine 23 The variable speed AC motor used in the MK-5 machine has poor performance under load, and the speed of the conveyor belt is influenced by the load due to fish. Sometimes, operator needs to remove jammed fish in order to release the conveyor. The variable speed DC motor used in the new design performs well in maintaining a constant speed under varying load conditions. T h e Sloped Conveyor Section Figure 2.9 presents two diagrams for design of the sloped section of the conveyor, Figure 2.9(a) illustrates a conveyor belt with flanges and Figure 2.9(b) illustrates a conveyor belt with rubber cylinders. Note that a denotes the slope angle of the conveyor belt. The maximum speed of the conveyor belt (K) is drawn as a horizontal dotted line on the diagrams. Another dotted line, named \"On-Plate\", corresponds to Figure 2.8 and serves as a reference. When fish are loaded on the inclined conveyor belt, the flanges ( or rubber cylinders) maintain a single layer ( i.e., the bottom layer) of fish in the conveyor. All other fish (i.e., that in the top layer) will slide down and will be separated from the bottom layer. The mark * on the diagrams corresponds to the data sampled from the tests for the bottom layer. One dotted line through the origin of the coordinate axes, has been drawn based on the data and named \"Bottom-Layer\". Similarly, the mark + denotes the data from test for the top layer, and another dotted line, named \"Top-Layer\" is drawn based on them and the original point. The basic rule to draw these two dotted lines is that most of data for the top and the bottom layers from the test should be located outside the area between these two dotted lines. Note that the angle axis is drawn to a sine scale. It is easy to see from the diagrams in Figure 2.9 that the maximum conveyor speed (Vc) is at the center between slipping speeds of the top layer and the bottom layer of fish, and this occurs at about 30° of the slope angle a. When a continues to increase beyond Chapter 2. Design Development of the Spray Marking Machine 24 30\", the two lines of Vc and bottom-layer become closer and will merge at an angle of about 45° of a. In other words, the sliding down speed of the bottom-layer fish will increase and become closer to the maximum speed of the conveyor (Vc). A mechanical limit for the range of the conveyor slope angle ( Q ) is set at 33°, as shawn by the dotted line, \"Mechanical Limit\", in the diagrams of Figure 2.9. A dotted area, enclosed by the four lines Vc, the Mechanical Limit, the Top-Layer and the Bottom-Layer, is the operating area. For the single layer delivery, a suggested value of the conveyor speed can be found based on a given angle a, or alternatively, the angle a may be determined on the basis of a given conveyor speed specification. Discussion The new design for the conveyor attempts to overcome the overlapping problem that exists in the MK-5 machine. The conveyor speed and the angle of the sloped section are discussed above in terms of the maximum marking speed and the different viscous damping coefficients of the mucus on the fish body in the bottom layer and the top layer. The adjustable nature of the conveyor speed and the slope angle is used to compensate for any variation due to changes of fish size, damping force, and marking speed. Using a matrix of rubber cylinders (or fingers) which are about 12 Tnm high, on the conveyor, results in a considerably better single-layer performance than what is possible when flanged conveyors are used. For coded marking in the future, however, straight flanges are needed to keep fish at a certain fixed orientation as they pass through the spray marking region. The marked fish are live juveniles which may try to jump on the conveyor. Fish jumping makes the design of the conveyor much difficult than initially anticipated. When fish move along the sloped section of the conveyor, a fish sometimes jumps on top of another just before entering the region which has a top guard mesh above the conveyor Chapter 2. Design Development of the Spray Marking Machine 25 A V (mm/s) 600 500 400 300 200 100 - - - - V On-Plate .J. r legend: + —- vt * - — Vb Vc Mechaoical Limit -%' If ! + iJop-Lay^r\" I ]B6tt6m-l_feyer I . - -T * / I a T o T T T 10\" 20\" (a) With Flanges 30\" 40^ Angle (Sine) '' V (mm/s) 700 600 500 400 300 200 100 legend: + -— vt * - — Vb On-piate / / I Vc Mechanicail •*• Limit \\ ^ + + Top-L9yer Bottom-Layer a O T 10\" 20\" (b) With Rubber Cylinders 30\" 40^ Angle (Sine) Figure 2.9: Speed-Slope Data for Designing the Sloped Conveyor Segment Chapter 2. Design Development of the Spray Marking Machine 26 compartments. This can happen quite often. If the conveyor and the guard are made of hard material and the clearance between them only allows one layer of fish to go through, the fish on top may be wounded or killed once it gets caught underneath the guard. On the other hand, a large clearance means that fish could go through the machine safely although the overlapping problem would remain. So the clearance between the guard and the conveyor is a critical parameter, and some compromise must be made in selecting this parameter. 2.3.2 The Feede r The feeder holds and delivers fish onto the conveyor through its exit slot. In the new design, an integrated feeder and conveyor system is utilized because the configuration of the feeder is another key part on influencing the orientation of the fish and the overlapping problem. The feeder in the MK-5 machine, as shown in Figure 1.1, has a triangular shaped cross section. The bottom wall of the feeder continues up to the slot, and the size of the slot is used to regulate the feeding speed. From previous tests, we found that the fish rush to the slot and squeeze out into the conveyor. Some fish were damaged or killed during this process. Because of variation in the fish quantity and fish size in the feeder, the feeding speed could not be well regulated. Fish jamming at the slot and overlapping on the conveyor were common occurences in the old machine, which resulted in a low efficiency of the machine and a poor quality of marking. A zig-zag channel is constructed as part of the newly designed feeder. With a sta-tistical analysis of the fish random behavior of the out-of-water, this zig-zag channel is studied, to provide means for overcoming the overlapping problem that existed in the MK-5 machine. The constrained orientation of fish, combined with the feature of the Chapter 2. Design Development of the Spray Marking Machine 27 Zigzag Shaped Channel Guard / / / / / / / / / / / Ground 7 7 ^ Exit Figure 2.10: The Feeder with a Zig-zag Channel flanged conveyor belt, is proposed as the useful stepping-stone for the future marking of a code. T h e Feeding Speed Regulation As shown in Figure 2.10, the zig-zag channel of the feeder has two stages, and each of them has a flow-deflecting edge. When fish come down to the zig-zag channel from the top of the feeder, they change their moving direction at each flow-deflecting edge with a lose of monaentum. In this manner, the speed of the fish that is coming out the feeder will be regulated to some extent. Also, the randomly arranged fish at the entrance of the feeder will tend to arrange in a more organized manner at the exit, due to the two stages of deflection. The regulating principle can be described using Figure 2.10, and considered under the following cases: Chapter 2. Design Development of the Spray Marking Machine 28 Case 1: If one fish has been put in the feeder, the fish will slip down and will accelerate due to gravity. When it gets to the corner of Stage 1, the fish will collide with the deflecting surface and come to a virtual rest. It will then accelerate again in the new direction, as a result of gravity. Fish will be stopped again upon rearching the deflecting surface of Stage 2. The exit speed of fish from the feeder will depend on factor such as, potential energy of fish at the top entrance of the feeder (depends on elevation), friction characteristics of the sliding surfaces, and the energy loss due to the collisions. Another interaction occurs as the fish enter the conveyor. This will depend on the conveyor speed, relative orientation of a fish with respect to the conveyor, and the material characters and friction of the conveyor belt and the compartments. The final speed is, therefore, constrained to some extent. Case 2: If several fish are placed in the feeder, they will reach the corner of Stage 1 at approximately the same time. When one fish gets to the corner, it stops and begins to slide away from the corner by gravity. This fish might be further accelerated due to impact with the fish that follow it. Similarly, after passing through Stage 2, fish will likely come out in a queue, in a certain speed range. Case 3: If a large bulk of fish is dropped into the feeder, the fish will likely reach the deflector plate of Stage 1 in a jam. The flow speed here is regulated by the squeezing action of fish that are present above this corner. The squeezing pressure depends on the height of the fish bulk inside the feeder. It is understood that fish are subjected to shear force at the slot of the deflecting edge due to this pressure. Similarly, fish will jam up again at the opening of the second deflection surface (Stage 2). Note that the fish flow rate in Stage 2 is less than that in Stage 1 because the clearance of the slot of Stage 2 is smaller than that of Stage 1 according to the design. When Chapter 2. Design Development of the Spray Marking Machine 29 the piled-up fish in Stage 2 reach the slot of Stage 1, a inverse pressure will build up against the motion from Stage 1, and the shear force on fish at the slot of Stage 1 will decrease. Some pressure at Stage 2 is built up from Stage 1. It is transmitted right to the wall of the channel which will slightly increases the fish flow rate at Stage 2 in comparison with that in Case 2. Finally, the general feeding rate is regulated through the entire feeding channel and fish are protected from experiencing a large shear force at the slot edges. The feeder can be designed in several ways to reach the goal of regulating the fish flow. By taking into consideration of the conveyor speed, the overlapping problem will be eliminated in this new design. T h e Fish Orientation The orientation of the fish as they reach the conveyor is another important factor. From the discussion regarding the spray gun system, it was clear that the multi-gun system is able to cover the entire width of the conveyor belt during marking. All the fish on the conveyor will be marked by this arrangement and their orientation on the conveyor is not important. However, if the fish can be placed across conveyor at right angles to the conveyor motion, then assuming that the length of fish ranges between the half width and the whole width of the conveyor belt, all the fish on the conveyor can be marked with just one or two gun system. So good orientation of fish can help the marking quality. Further more, if the marking region has a slotted cover with diff\"erent spacings and colors across the conveyor, it will enable the application of an entire code, instead of a simple binary mark. A further new design can have this potential for marking with a code. Once the fish are dropped into the feeder, they will slide down along the bottom side of the feeder into the zig-zag channel as shown in Figure 2.11. Before getting into the Chapter 2. Design Development of the Spray Marking Machine 30 Figure 2.11: The Zig-zag Channel zig-zag channel, the fish tend to have a randomly distributed orientation. After entering into the zig-zag channel, fish will undergo a regulated motion as described previously and will tend to orient themself in a regular manner. As a result, more fish will end up oriented across the conveyor, perpendicular to the conveyor motion. The parameters of the two-stage zig-zag channel include: s = the distance from Stage 1 to Stage 2; (p = the angle between the bottom plate of the feeder and the ground; Ci,C2 = the slot clearances of Stage 1 and Stage 2, respectively; Mi,M2 — the orientation reading points of Stage 1 and Stage 2, respectively. As described above, the fish orientation is assumed to have a uniform distribution because of the random motion of fish. If only the axial direction of orientation of the fish body is considered without taking into account the difference between the fish heading direction and the fish tail direction, with respect to the perpendicular direction to the conveyor motion, the fish orientation distribution can be drawn as in Figure 2.12. The variable 6, which denotes the fish orientation with respect to the across direction, ranges Chapter 2. Design Development of the Spray Marking Machine 31 f(e) 1/180 90 90^ Figure 2.12: The Uniform Distribution of Fish Orientation from —90° to 90°. It is a reasonable assumption not only because the distribution of fish orientation is reasonably symmetric along both the across direction and the moving direction of the conveyor, but also because the heading towards one side of the conveyor or towards the opposite side will not be considered differently for the purpose of marking. When the fish passing through the zig-zag channel of the feeder, the distribution of the fish orientation will be biased in a particular way. A series of experiments have been carried out to study this change in distribution. The experimental setup, the fish size and the results for a one-stage or two-stage channel are given in Appendix A. To measure degree of bias in the distribution of fish orientation, a distribution index p has been defined as njid)\\s\\nd\\dd J^lof (6) cos Ode (2.9) where f(6) = the distribution of fish orientation. It is understood that the distribution index p is the ratio of the area of the distribution, one biased towards the central region (6 = 0°) and the other biased towards the end Chapter 2. Design Development of the Spray Marking Machine 32 regions (6 — ±90°). Specifically, the denominator of equation ( 2.9) gives the bias of the distribution towards the direction ^ = 0° and the numerator provides the bias towards the directions 9 = ±90°. Hence, the case of /> = 1.0 corresponds to an evenly random distributed orientation. If p is less than 1.0, more fish tend to lay along the direction ^ = 0 than that in the across direction, and vice versa. For example, a case oi p = 0.2 could be interpreted as 10 fish on the average tend to lay along the 0° direction and 2 fish on the average tend to lay in the across direction. For the samples of fish in the experiments, define an orientation distribution index as Pd = ^= j^7 — (2.10) E„=lCOS0„ Where N = the sample size (number of fish measured); n = the denoting number of an individual fish. The fish orientation distributions of the tests and the corresponding values of pd are shown in Figure 2.13. From equation ( 2.9), we kown that p — 1.0 corresponds to a uniform distribution. When fish pass through the zig-zag channel, p becomes smaller, indicating that they increasingly tend to orient across the conveyor. It is clear that the smaller the p is, the larger the bias of the distribution towards ^ = 0°. According to the discussion in the previous section, the ideal value for p is 0, which produces the best marking orientation, i.e., all the fish would be oriented across the conveyor. The depth of the biased distribution is influenced by parameters s and of the chan-nel, shown in Figure 2.11. From experiments Zl , Z2 and Z3, in which all the experimental conditions are the same except for the parameter s, we know that the smaller the s is. Chapter 2. Design Development of the Spray Marking Machine 33 026 020 1 g OIS i oio-^BL ffil 00 •60 •40 30 20 10 04 • 0 3 -1 1 5 02 01 - r-p-r-i-H ri-n-i-n , -90 -zo 80 120 Z1M1: N=262 0=37.826° p=0.5550 Min. -90° Max.+90° Mean 6.641 ° Z1M2: N=244 0=29.181° p=0.3384 Min. -90° Max. +90 ° Mean 2.992 ° 026 020 I ^ 016 -006 ^ Mn 30 00 0 6 n 04 p-ruTP IW eo •40 -90 -30 30 go Z2M1: N=158 0=34.960° p=0.4762 Min. -90° Max.+90° Mean 3.291° Z2M2: N=203 0=30.086° p=0.3085 Min. -90° Max. +90° Mean 2.167° Figure 2.13: Test Results on Fish Orientation Distribution Chapter 2. Design Development of the Spray Marking Machine 34 020-S [ 006 ' - T T I M -r r i IfTD , -30 -20 -10 020 -20 eo 120 jn fl. •20 m. -90 -eo -10 30 70 110 Z3M1: N=221 0=41.246° pa=0.6505 Min. -90° Max.+90° Mean -0.407° Z3M2: N=202 0=39.034° ft,=0.5294 Min. -90° Max. +90 ° Mean -5.099^ 028-1 020-018 • 010-0 0 6 ' M^ -] f^ 20 026 n 020 ^ ots ^ 006 -90 -30 -90 -20 60 120 r-eo •60 •40 •30 h20 • 10 Z4M1: N=180 0=39.528° pb=0.5658 Min. -90° Max.+90° Mean 4.722° Z4M2: N=223 0=33.810° pt,=0.4368 Min. -90° Max.+90° Mean -0.762 Figure 2.13 (cont'd) Chapter 2. Design Development of the Spray Marking Machine 35 the smaller the p^ at Stage 2. This is because a biased distribution after Stage 1 will try to return to a uniform distribution when fish travel through the distance s before being biased again at Stage 2. Similarly, the experiments Zl and Z4 use the same experimental conditions except for (f> and Ci. The orientation distribution at Stage 2 is biased much more in experiment Z2 than experiment Z4. It is clear that a sharp turn in the flow direction of fish enhances the stage biasing function. Another experiment considers the jumping (or random tossing or moving) effect of live fish. Due to the biasing effect of the zig-zag channels, the distribution index p will be reduced from 1.0 (uniform distribution) to a smaller positive value (biased distribution). If the fish are initially subjected to a biased distribution, they will eventually return to the uniform distribution as a result of the random movement of the free fish. Sixteen fish were used in this experiment. Every fish was tested using the following procedure: First, a fish was oriented at 0° on a horizontal plate, which is covered by a net to increase the surface roughness and simulate the marking machine. The orientation of the moving and tossing fish is then recorded using a video camera. Figure 2.14 illustrates the change in orientation over time for each fish sample. Based on this data, the dynamic distribution index Pd{t) is calculated. Figure 2.15 illustrates the function pdit) which increases with time to a steady value of approximately 1. The irregular fluctuation of ripples in Figure 2.15 could be decreased by increasing the number of fish in the experimental sample. Suppose that pdit) is an exponential function of time, given by pd{t) = 1 — e~~, which is drawn as a dashed line in Figure 2.15. The time constant ( T ) of the dynamic distribution index is about 0.2 sec. This value of T is specific to the experiment and to the fish that were used, and will vary if experimental conditions and the fish sample are changed. Chapter 2. Design Development of the Spray Marking Machine 36 100 -en $ o > •a, c o c 6 -50 -100 -0.00 0.10 0.40 0.50 0.20 0.30 Time (sec) Figure 2.14: Experimental Results on the Dynamic Orientation of Fish 0.60 0.00 0.00 0.10 0.20 0.30 0.40 Time (sec) 0.50 0.60 Figure 2.15: Experimentally Determined Dynamic Distribution Index Pd{t) Chapter 2. Design Development of the Spray Marking Machine 37 2.3.3 The Feeder and Conveyor S y s t e m The design of the feeder and conveyor system focuses on the specification of the operating parameters between the conveyor and the feeder, as discussed above. There are two important parameters: flow rate, and fish orientation in the marking process. The flow rate in the feeder corresponds to the number of fish that leave the feeder, at the exit, per unit time. This rate, which is influenced by fish size, and the elevation (length and slope) of the feeder unit, may be regulated by the slot at the exit of the feeder, especially for a zig-zag channel. The critical delivery rate of the conveyor is defined as the number of fish per unit t ime that are delivered by the conveyor belt when fish are located on the belt side by side without overlapping. It depends on the fish spacing and the speed of the conveyor speed. The conveyor speed is determined by calibrating the DC motor that drives the conveyor system. Meanwhile, the angle of inclination of the sloped-up section of the conveyor is set according to the speed and the characteristics of the fish as shown in Figure 2.9. To eliminate the overlapping problem, the critical delivery rate of the conveyor must be larger than or at least equal to the flow rate of the feeder. The fish orientation is another important factor in the design of the machine. From the discussion in the last section, when the fish go through the zig-zag channel, the distribution of the fish orientation changes from uniform distribution (p — 1) to a biased distribution (p = pi < 1) immediately after passing through Stage 1, then changes from this biased distribution [p = pi) to a weakened biased distribution (p = p[, pi < p[ < I) when moving from Stage 1 to Stage 2 because of the random movements of fish. Similarly, the distribution of the fish orientation is biased for the second time to another biased distribution (p = p2 < p\\ and p\\) when passing through Stage 2, and it changes to another weakened distribution [p = p'2) at the time the fish finally reaches the conveyor. Chapter 2. Design Development of the Spray Marking Machine 38 The surface of the conveyor belt is not a open flat area due to the presence of the matrix of rubber cylinders or regularly spaced flanges. The choice of the material used for various components of the conveyor influences the fish orientation in diff'erent ways. Flanges are usually mounted right across the direction of the conveyor motion, and these tend to move fish along with them during transportation. Considering the effect of random movements of fish, the fish orientation distribution will change to a particular index value (pf), where 0 < /)/ < 1. This index value (pf) is affected by the height of flanges and the spacing between the flanges. For the particular conveyor belt used in the experiment with a slope of 22° from the ground, pf was found to be 0.34 when the fish were densely placed in one layer, and pf was found to be as low as 0.20 when the fish density was low, so that there were some empty compartments in the conveyor. The rubber cylinders on the conveyor belt keep the fish orientation more or less unchanged during transportation. Actually, the fish orientation usually changes slowly with time, and given ample time the biased fish orientation distribution would eventually change back to the uniform distribution (p =^ 1). The time constant (T^) of this dynamic orientation distribution could be as small as 0.4 sec in some cases, but is larger than T for an open belt without compartments or a matrix of orientation cylinders, as shown in Figure 2.15. The less comfortable the fish are on the conveyor, the smaller the value of T under semi-constrained conditions. A guard made of metal mesh keeps fish from jumping during delivery on the conveyor. Suppose that the fish orientation distribution does not change as the fish enter the guard region. Then the distribution index (/)) will remain unchanged while the fish are being transported under the guard. The diagram of a representative distribution index as a function of t ime is shown in Figure 2.16. This represents the entire process from the time the fish enter the feeder until they leave the conveyor after marking (exit of the guard region). The dashed line Chapter 2. Design Development of the Spray Marking Machine 39 P(t) 1.0 0.5-i^' fi— PT/ Zig-zag Shaped Sloped Channel Section of Guard Region Conveyor Segment Staged Tc ^ ^ Stage 2 1^ / J -t 0.0 1.0 2.0 3.0(sec) Figure 2.16: Representative Distribution Index for a Design Prototype represents a conveyor belt with flanges, and the dotted-dashed line for a conveyor with a matrix of rubber cylinders. The diagram is a good approximation for the real situation. The biasing of the fish orientation is intended to achieve a distribution index (p) as close to zero as possible during the marking process. From the discussion above, we know, in general, that the distribution index for the marking process can be lowered by shortening the processing time from Stage 1 to the entrance of the guard region. It must be mentioned that the irregular tossing, wiggling, and sometimes violent motions of live fish out of water can cause significant difficulties in the design of the feeder and conveyor system. Using a mild anaesthetic for fish in the holding tank, prior to marking, is a conventional method to reduce this problem and associated stress on and damage to fish during the marking. However, this method has not been used in the experiments which we carried out, as we wished to consider the natural behavior of fish. If an anaesthetic were used, the experimental data and parameters could be modified to take this factor into account. Furthermore, the design of this system would be simpler for anaesthetized fish. It is not recommended to use these chemicals, however, in view of the potentially negative environmental impact of such procedures and Federal regulations on Chapter 2. Design Development of the Spray Marking Machine 40 the use of anaesthetics and the time of release following treatment. 2.4 The P igment Recycl ing System During marking, only a small fraction of the sprayed fluorescent pigment becomes em-bedded in the fish body, and this would represent the useful mark. The most part of the pigment simply sticks to the conveyor belt and to the surfaces of other parts of the machine, and eventually drops to the bottom tray of the machine. A small fraction of the pigment may escape the marking machine into the air as dust, as the machine is charged, and finally will fall onto the ground. This is minimal when an emulsion instead of the dry powder is used. This pigment emulsion, developed for the spray gun system, as described in section 2.2, makes recycling of the pigment possible. A pigment recycling system is designed to collect the used pigment, to reduce the wastage of the pigment material and to keep the working environment including the water system and the fish pond clean. It is environmentally friendly and also less hazardous to operators. The pigment recycling system in the newly designed marking machine, as shown in Figure 2.17, includes a mixer, a water-spray cleaner, a water tank and a number of hoses. These features and the working procedures are described below: 1). The Mixer The mixer is used to mix the dry fluorescent pigment powder with water, thereby producing the pigment emulsion for spray marking. The ratio of the pigment to water in the emulsion is controlled by the operator. When the machine is running, the recycled emulsion, that leaves the water tank, can be reused in the mixer. The recycled emulsion, however, has been diluted by water carried by the fish and also from the spray wash system. Its concentration is lower than what is appropriate for the emulsion. So the dry pigment powder needs to be added Chapter 2. Design Development of the Spray Marking Machine 41 Compressed Air Emulsion Pump Water Pump Figure 2.17: A Schematic Diagram of the Pigment Recycling System Chapter 2. Design Development of the Spray Marking Machine 42 into the mixer in order to maintain the emulsion concentration at the proper level. The method of determining the amount of dry powder to be added in this manner is discussed in Chapter 5, under future work. 2). The Spray Wash System The spray wash system has two parts. The first part is the water pump with a built-in motor and the second part is the shower head attached to a hose. Water from the upper layer of the water tank is pumped through the hose to the shower head, and it is then sprayed on to the exit region of the conveyor and the mesh guide on which the marked fish slide down from the conveyor as they leave the machine. The sprayed pigment sticking onto the conveyor and the excessive pigment on the fish body can be washed off in this manner, and be collected by the retaining tank at the bottom of the machine. 3). The Retaining Tank The retaining tank is made of PVC sheets and is used to collect the pigment which drips down from the other parts of the machine. It also supplies water for the spray wash system. After dripping into the tank, the pigment in the emulsion settles on the bottom because of its higher mass density, and it may be reclaimed from the lower outlet (Figure 2.17). The upper layer of the reclaimed suspension has a lower concentration of pigment, and is almost clear. This water is used in the spray wash system. The water carried into the machine by the fish from outside will make the water level gradually rise inside the retaining tank. The overflow outlet is used to maintain a constant water level in the tank. The original spray marking technique has a very high marking speed, but it makes poor use of the pigment material. Because only a small portion of the total sprayed pigment is retained in the useful mark of the fish, the old design may waste over 95% of the pigment. With the new design, which has a pigment recycling system, the usage rate Chapter 2. Design Development of the Spray Marking Machine 43 of the pigment materials will be improved considerably. For further study, mass marking experiments have to be carried out with a large batch of fish, to quantify the efficiency of the new system. Chapter 3 Mark Retent ion Study 3.1 Introduction The overall design of the automated spray marking machine for fish was developed and studied in Chapter 2. From the spray marking tests we conducted in the laboratory and at hatchery sites, it was observed that the newly developed machine performs much better than the model MK-5 machine. To investigate the influence of the design parameters such as the air pressure supplied to the spray gun system, feeding rate of fish into the machine, and pigment particle concentration on the body of a marked fish on the performance of the new machine, a technique based on image processing for measuring the mark retention of the sprayed fish has been developed. This method is used to eliminate the uncertainty in mark perception of human observers as they evaluate the quality of marks in a batch of spray marked fish. Note that manual observation by experts has been used so far in determining the mark quality. Measurement of the short-term and the long-tei^.i mark retentions is useful in eval-uating the performance of the newly designed, automated spray marking machine. If any defects or malfunctions of the marking machine are found in this manner, the corre-sponding components of the machine could be improved through corrective action, such as design modification. This checking and redesign loop can be repeated until the re-quired performance is achieved. The adjustment and fine tuning will be done at the end. It follows that the study of the mark retention plays an important role in the design of 44 Chapter 3. Mark Retention Study 45 the new automated spray marking machine. As discussed before, the marking method used in the machine consists of spraying a fish with a suitable fluorescent pigment which becomes lodged in the scales of the fish. Some pigment particles simply stick to the mucous film on the body of the fish and are washed away in a relatively short time. Also, as a result of turbulence and abrasion as the fish interacts with their environment, further particle loss will occur. Furthermore, the pigment marks on the fish body change from a continuous patch to a collection of discrete particle spots, as the fish grows. All of these factors influence the mark retention, and will be studied using image processing techniques. A system for mark retention study is developed in this chapter. The hardware and the methods used, the basic image processing techniques, indices of mark retention, and the pigment mark recovery will be discussed separately in sequence. 3.2 Sys tem Hardware and Techniques The assessment method uses a commercial computer vision system with associated soft-ware (SHARP GPB and InCard), and an ultraviolet lamp, (model B-IOOA). As discussed in Chapter 2, the fish are sprayed with an emulsion of fluorescent pigment of diameter 50 to 350 fim and water, and transferred into a retention tank. Subsequently, at fixed inter-vals of time (1 month) the marked fish are examined as follows: First, a fish is placed on a green background (back lit) and illuminated by the ultraviolet source. The test region is enclosed and no other ambient fighting is used. The green image channel is used to measure the features of the fish body, and similarly the fluorescent grit particles on the body are measured using the red image channel. The set-up of the device is schematically shown in Figure 3.18. The mark retention study system is further discussed, with details of the techniques, in the following sections [11]. Chapter 3. Mark Retention Study Green Channel CCD CameraJ 46 486-Based Computer Computer Monitor Image Monitor Figure 3.18: The Block Diagram of the System for Mark Retention Evaluation 3.3 Basic Image Process ing For a reliable and efficient method of evaluating the spray marking system developed in our laboratory, basic image processing techniques, such as histogram and thresholding techniques [12], are employed. The lighting arrangement plays an important role in cap-turing the image. Good lighting simplifies the image processing procedure and enhances the quality of the processing results. A histogram of pixel brightness of an image can be defined as K = fib) (3.11) w here b — the pixel brightness, ranged from 0 to 255 in integer; Chapter 3. Mark Retention Study 47 K = the pixel quantity; f[b) = the pixel density with pixel brightness. The total number of pixels in a sampled image is 512x480 or 245760. Therefore equation 255 ^ / ( 6 ) = 245760 6=0 must exist for all images. And all of diagrams are drawn in logarithm for the pixel density due to this large number. Figure 3.19 shows the histogram of pixel brightness for a fish placed on a white background with illumination using standard overhead room lighting. There are three main peeks {Pi, P2, P3) and two main valleys (V i^, V2) in the histogram. It is evident that under standard lighting the fish body has great variation in grey-scale values, the pixel density of the fish body is widely distributed with brightness, and that the test platform is flat and has uniform surface conditions. So the pixel density for the background in the image is concentrated around a grey-scale value that corresponds to the large middle peak (P2) in Figure 3.19. Other two peaks (PijPa) are obviously dominated by the fish body in the original image. To calculate the area of the fish body region within the image, select two threshold values corresponding to the valley points Vi and V2 in Figure 3.19 and segment the image into 3 regions, with regions A^ and A3 representing the fish and region A2 representing the background. The number of pixels in regions Ai and A3 of the image is a coarse approximation of the area of the fish body. Since both pixel distributions of the back-ground and the fish can overlap, this will affect the accuracy. Specifically, there are two drawbacks to this approximation. First, the area value is very inaccurate; and second, two thresholds are needed to separate region A2 from Ai and A3. Because structured lighting can be used to modify the shape and the peak location Chapter 3. Mark Retention Study 48 E 2 3.0 -2,0 0.0 100 150 Pixel Brightness b Figure 3.19: The Histogram of the Image under Ambient Light of the pixel intensity distributions of the background and the object, it is possible to overcome the overlapping problem of the distributions. As a result, only one threshold would be needed to separate the object from the background in the image, and the area of the object could be obtained accurately. The lighting arrangement described in the previous section is used to measure the level of the pigment particle retention of the fish mark. To demonstrate this method, a green image and a red image are captured through the green channel and the red channel respectively, and the histograms are drawn in Figure 3.20 and 3.21. There are also three peaks (Pi ,P2,^3) and two valleys (VjjVa) in the diagram of Figure 3.20. Unlike Figure 3.19, the peaks Pi and P2 in Figure 3.20 correspond to the fish, and peak P3 corresponds to the background. The green background (back lit) has the highest brightness on the green channel, and the pixel distributions of the fish and the background do not overlap. One threshold is needed at valley V2 to extract the Chapter 3. Mark Retention Study 49 E 5.0 -4 0 -3 0 2.0 -10 -0 0 ' — I 1—I 1 1 100 150 Pixel Brightness b Figure 3.20: The Histogram of the Image via the Green Channel two regions {Bi, B2) from the whole image, in Figure 3.20. The area of the fish body is estimated as the number of pixels in region Bj . This area approximation is more accurate than the estimate based on Ai and A^ in Figure 3.19. Similarly, there are two peaks (Pi,P2) in the histogram of the red channel image, cis shown in Figure 3.21. Since the fluorescent pigment mark on fish body under the ultraviolet light is red, the brightness of the mark is higher than that of the dark fish body or of the green background. A threshold is set at valley Vi to extract the two regions (B^jB^) in the diagram. The area of the spray mark on the fish body is approximated by the number of pixels in the region B4. 3.4 Two Indices of Mark Retention Based on the image processing principles discussed thus far in this chapter, two indices are proposed to express the pigment particle retention of a marked fish. The idea here Chapter 3. Mark Retention Study 50 I 100 150 Pixel Brightness b Figure 3.21: The Histogram of the Image via the Red Channel is that effective spray marking should result in a large number of particles being lodged in the scales over the entire surface of a fish. First, note that an indication of the light energy can be given as the sum of the pixel intensities of all the particles on the surface of the fish. The division by the fish body area is necessary to account for the surface area variation among fish. Denote the light energy of the pigment mark on the body of a particular fish (the n-th fish) as: where Ay is a region in the histogram in Figure 3.21. Since the long term effectiveness of the spray marking technique has to be checked periodically as the marked fish grow, an index for the retention measurement of a batch of fish may be expressed as: -'* n = l -^n (3.12) Chapter 3. Mark Retention Study 51 where Rb = the index of retention in a batch of fish; N = the batch size; Sa = ' ^ — - , the average body area of fish in the batch; En = the light energy of the pigment mark on the n-th fish; S„ = J2beBi /(^)) the individual body area of the n-th fish; Bi = a region in the histogram in Figure 3.20. This index takes into account the average growth of a batch of N fish and allows for long term mortality within the batch. Another index that is practical and simple is provided for the particle retention measurement of an individual fish: R^=^ (3.13) where R, — the index of fish mark retention; Sp = ^beB^ /(^)) the area of the pigment particle mark on the fish body; Sf — J2beBi /(^)> the area of the fish body. Bi — a regions in the histogram in Figure 3.20; i?4 = a regions in the histogram in Figure 3.21. The fluorescent mark on the fish body is made up of pigment particles, which may be divided into two groups, overlapping and non-overlapping. Since the number of pigment particles will not increase as a fish grows, the index in equation ( 3.13) will decrease Chapter 3. Mark Retention Study 52 gradually with time even without particle loss due to fish activity, including turbulence and abrasion. Both of these indices have been implemented on the image processing computer sys-tem. Although there are still a number of factors, such as the fish growth, the loss of the mark pigment and the water blobs on the testing board, affecting these indices in a complex manner, the methods have eliminated the subjective uncertainties in human assessment of mark retention. 3.5 P igment Mark Recovery Immediately after marking a fish, the pigment particles are concentrated in some regions of the fish body and the mark can be treated as continuous. Subsequently, in the short term, some particles are lost from the fish body. With the growth of the fish, the marked regions will become less dense and will eventually appear as a set of discrete dots of pigment particles. With further loss of pigment particles and further growth of the fish, the distance between the particles in the same region increases. From this observation we conclude that by connecting a particle with its neighboring particles inside each marked area, a region with a similar configuration to that of the original mark can be obtained, assuming uniform growth of fish, and no loss of particles during the period. In this manner, the entire pigment mark in the image may be reconstructed. Consequently, the index for particle retention of an individual fish may be improved, as equation ( 3.14), by using the total area of the marked regions instead of the total area of the pigment particles in equation ( 3.13). Sharp GPB standard image processing functions, such as binary \"dilation\" and \"ero-sion\", can be used to connect the neighboring pixels inside a marked area for a binary image. A parameter block, associated with these two GPB functions, has a format as Chapter 3. Mark Retention Study 53 A D G B E H C F 1 Figure 3.22: The Parameter Block shown in Figure 3.22. The weights of all of pixels in the meshes are either 0 or 1. For example, when using the \"dilation\" function, the brightness value of each pixel in the original image will be factored into a revised brightness value for the nine pixels in a new image according to the parameter block of Figure 3.22, with defined weights. The center position which is denoted as pixel \"E\", which is weighted in the new image, correspons to the position of the pixel in the original image that was stretched. On the contrary, the \"erosion\" function with the 3 x 3 parameter block of Figure 3.22, will cause a pixel in the new, transformed image to be set high only when the corresponding parameter block pixels in the original image match the pattern defined in the parameter block. There are three symmetric patterns with \" 1 \" at the center, as shown in Figure 3.23, defining the parameter block for our purpose. In the figure, (a) is a square pattern, (b) is a \" + \" pattern and (c) is a \" x \" pattern. One example of image processing with the \"dilation\" and \"erosion\" functions is shown in Figure 3.24. On this figure, (a) is an image with three connected pixels with a high brightness, and (b) is the square pattern of the parameter block used for the \"dilation\" and \"erosion\" operations in image processing. Note that (c) is the resulting image with eleven pixels. It maybe easily noted that image (a) can be spread out (dilated) into image (c) through the \"dilation\" function with pattern (b). Conversely, image (c) can be shrunk (eroded) to image (a) using the \"erosion\" function with pattern (b). Chapter 3. Mark Retention Study 54 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 (a) (b) (c) Figure 3.23: The Three Symmetric Parameter Blocks 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 (a) (b) (c) Figure 3.24: An Example of Binary Image Processing in Dilation and Erosion For a brief analysis in the following section, suppose that a small square block stands for one pixel in the image, where the marked square stands for a pixel with a high brightness, (instead of \" 1 \" used above), and the empty (or blank) square stands for pixel with a low brightness. There are two methods to make connections: one is called the Square Dilation and Square Erosion (SDSE) method, and the other one is called the Square Dilation and \"Plus\" Erosion (SDPE) method. Both these methods will be studied in the following section. 3.5.1 The Square Dilation and Square Erosion Method In the square dilation and erosion method, we use the dilation function with a square pattern for the parameter block repeatedly to dilate an image k times and the erosion Chapter 3. Mark Retention Study 55 (a) (b) (c) (d) (e) Figure 3.25: An Example of Image Processing with SDSE function is then used with the same pattern for the parameter block to erode the image successively for another k times. As a result, the nearby scattered bright pixels in the original image will be connected into a combined area. The key analyses are shown in Appendix B, for the case of A; = 2. The leftmost column shown in Appendix B denotes the case number. The next three columns from left to right give the original image, the final image and the increased number of pixels with a high brightness, respectively. The extended images, first dilated from the original ones and subsequently eroded to the final ones, are not drawn in Appendix B. The same procedure is applied to the example shown in Figure 3.25. The image in (a) is the original image. The case of one dilation is given by (b), the case of two dilations is given by (c), the case of one erosion is shown in (d) and the case of two erosions is shown in (e). It is identical to the Case 3 of Appendix B. Summarizing the results from Appendix B, we obtain two rules for applying the dilation-erosion operation: Rule 1: Two bright pixels in the original image will be connected by bright pixels if they are in the same line, i.e. in the same column or in the same row, and the number of dark pixels between them has to be 2k or less. These dark pixels are called pseudo-bright pixels. Rule 2: Any two pixels, regardless of whether they are bright pixels or pseudo-bright Chapter 3. Mark Retention Study 66 pixels, will be connected if they are in the same line, i.e., in the same column or in the same row, and the number of the dark pixels between them has to be 2k or less. 3.5.2 The Square Dilation and Plus Erosion M e t h o d As in the case of square dilation and square erosion method, this method also applies the dilation function with the square pattern of the parameter block for k times to spread the image, and then applies the erosion function with a \"+\" pattern of the parameter block for further k times to shrink the image back to the original size. Through this process, the bright pixels in the original image that are near the central pixel of an area will also be connected into the area. The key analyses are shown in Appendix C, for the case k = 2. The table arrangement in this appendix is the same as that given in Appendix B. Suppose that n^ denotes the distance along a row between two bright pixels. It represents the number of pixel shifts between these two pixels. Similarly, suppose that Uc denote the distance along a column between two bright pixels. As before, it represents the number of pixel shifts between these two pixels in the column direction. Considering the cases of image processing given in Appendix C, the following rules can be stated: Rule 1: Two bright pixels will be connected if the sum of the distances n^ and ric, along the row direction and the column direction, is equal to or smaller than 2{k -(- 1), except in the cases: Uj. — 0,nc = 2{k + 1) and ric = Q,nr = 2{k + 1). Rule 1 can be explained using Figure 3.26. It is understood that any one of bright pixels inside the enclosed region will be connected to the bright pixel at the center with this method. The region is the so-called the connecting region of a bright pixel in the original image. More analyses with three and four pixels are shown in the latter part of Appendix C. From that , further two rules may be stated: Chapter 3. Mark Retention Study 57 5? KT-f^ VJ^' Figure 3.26: The Connecting Region of a Bright Pixel Rule 2; Two bright pixels cannot be connected directly if the sum of Ur and Uc is greater than 2(k + 1). However, they can be connected indirectly if any other bright pixels are inside the overlapped connecting regions of these two pixels. Rule 3: Three bright pixels will be connected into a solid triangle area if these three pixels are inside each other's connecting regions. In an experiment, the marks were observed three times from the same fish at intervals of one month, through the red channel of the camera, as shown in the first column of Figure 3.27. Using the SDSE and SDPE methods, the same marked region was recovered as a continuous areas on fish, from the three red-channel images in the right two columns of Figure 3.27. Two assumptions made are: (1) the fish body area increases uniformly as the fish grows; and (2) the density of the pigment particles along the boundary of the marked region is high enough for recovery of this boundary. The recovered marks from the original mark region and the diluted mark region on the fish body should have similar configurations, and should have similar area ratios with respect to the fish body. Once the area of the marked region is available, the index for the particle retention Chapter 3. Mark Retention Study 58 Red Channel Image of the Marked Region Recovered Marks (SDSE Method) (SDPE Method) Time (months) Figure 3.27: An Example of Pigment Mark Recovery Chapter 3. Mark Retention Study 59 Table 3.3: Values of the Mark Retention Index in the Duration(months) ret,nd rei:„,(SDSE) rei:„,(SDPE) 0 0,156 0.157 0.158 1 0.076 0.126 0.132 2 0.057 0.102 0.110 Ixperiment measurement of an individual fish, as given by equation ( 3.13), can be modified as ret' ind S^ Sf (3.14) where ret'^^j — the modified index for an individual fish; Sr = the area of the marked region on the fish body; Sf — the area of the fish body. Computed value of ret,nd and ret[^, corresponding to the experimental results of Figure 3.27, are given in Table 3.3. As seen from the results in Table 3.3, the values of ret[^ decrease over time. In practice, the particles along the boundaries of the marked regions may carry more information than those inside the marked regions. It follows that the loss of boundary particles may cause difficulties in recovering the configuration of a marked region. 3.6 Discussion Particle retention measurement using image processing techniques has been studied. It was found that since the resolution of the computer vision system is much lower than that of a human eye, some isolated particles are not detected by the camera. Standard Chapter 3. Mark Retention Study 60 image processing functions, such as dilation and erosion, w/ere used to recover the pigment marks, and indices of fish mark retention have been proposed as the basis for an objective assessment of the design parameters for an automated spray marking machine for live fish. To recover and recognize the marks and the codes therein, further studies would be needed on considering such as thresholding, boundary analysis techniques and contour matching techniques. Chapter 4 Conclusions and Recommendat ions The work presented in this thesis concerned with the design and development of a spray marking system for live fish, with applications in fisheries management studies. Consid-erable improvements have been made in design of an automated spray marking machine. A prototype was built and has been tested at hatchery sites. It performed well, and the main drawbacks associated with the old MK-5 machine were eliminated in the new design. This chapter summarizes the key improvements of the machine, test results, and the system for mark retention studies. It also makes recommendations for future work. 4.1 Des ign of the Automated Spray Marking Machine In the design of the automated spray marking machine, one improved feature is the use of a pigment emulsion, instead of the dry powder that had been used in the old model. The pigment retention duration had been tested and it remained high over a period of one year when fish were observed regularly. With the use of the pigment emulsion, and due to further design improvements, the amount of pigment dust escaping from the marking machine is virtually negligible. Meanwhile, due to pigment emulsion it is possible to use the machine in damp operating conditions. Furthermore, recycling of the used pigment is possible, thereby minimizing the pigment wastage and the resulting environmental problems. With the multi-gun spray system, the compressed air and the pigment material are used at a higher efficiency and the marking quality is also improved. 61 Chapter 4. Conclusions and Recommendations 62 A zig-zag shaped feeding channel has been incorporated into the feeder. This system has been studied with live fish to determine the effect of this design in constraining the orientation distribution of fish as they reach the conveyor belt. Ideally, all fish should reach the conveyor belt in an across orientation, at right angles to the direction of the conveyor motion. The sloped section of the conveyor eliminates the fish overlapping problem during spray marking. The conveyor belt is compartmentalized with either flanges or a matrix of cylinder shaped rubber fingers. The latter has reduced the damage to fish due to clogging between the conveyor and the stationary parts of the machine, and has also reduced the discomfort on fish. The conveyor speed and the slope angle of the sloped section of the conveyor are adjustable. Through tests and design refinement, a good combination of feeder and conveyor parameters can be established, which result in a high marking quality. 4.2 Field Tes t s The prototype of the automated spray marking machine was built according to the design improvements described in this work, and was tested first at the Summerland Trout Hatchery on April 29, 1993. The total number of the fish (rainbow trout) marked was approximately 23,000. The mean length and the mean weight were 11.3 cm and 15.0 gm. For later identification of marked fish, the right ventral (fin) of each fish in the test batch was clipped. Fish were spray marked using the pigment emulsion at a concentration of 1:1 pigment/water ratio in volume, with the spray gun activating by liquid O2 as propellant at 100-130 psi. The net marking time was less than two hours not counting the delay in supplying fish to the machine. The marking speed was close to the design specification. The officers of the Fisheries Branch, Research and Development, Ministry of Envi-ronment Lands and Parks have commented that 'the spray marking machine is greatly Chapter 4. Conclusions and Recommendations 63 improved'. It is reported that only 15 fish died during marking, and most of these ca-sualties were due to accidental steping on them during loading into the hopper (feeder). Hence the mortality rate is very low (i.e., less than 0.07%), in comparison with the mor-tality rate of 6.2% of the old machine before improvement [10]. A good estimate of the retention could be based on the long term mark retention tests which will be conducted in 1994 as part of the annual creel census activities. The second field test was done on June 11, 1993 in Kaslo, BC with about 1.2 million kokanee. The right ventral fins of the test batch were clipped, for future reference and identification. Four months later, after being released into lakes, 94 fish were captured on October 14, 15 and 16, 1993. The total number of fish with good quality spray marks was found to be 92, and just 2 fish with right ventral clips did not carry an acceptable spray mark. It can be concluded from this data that the reliability of spray marking with this new prototype is quite high. 4.3 T h e Sys tem for Mark Retent ion Detect ion The system for mark retention detection was presented in Chapter 3. The purpose of the system is to eliminate the uncertainty factor introduces during manual assessment of mark retention. The system may be used in evaluation of the newly designed automated spray marking machine as well. For this purpose, a number of basic image processing techniques, such as histogram computation, thresholding techniques and dilation/erosion methods were employed. Two indices of mark retention were implemented into the system, in C language. One fish index is based on the light energy of the pixels of the pigment mark with respect to the area of the fish body. The second index is a ratio of the total area of the spray mark to the fish body area. The system provides numerical values for pigment mark retention, Chapter 4. Conclusions and Recommendations 64 based on these two indices. The techniques of recovering the original spray mark pattern after a growth and activity period of a fish ware investigated. Two methods SDSE and SDPE were developed based on the functions of dilation and erosion, and binary image processing using a commercially available image processing (Sharp GPB) Card. These two basic functions are popularly used to widen a line or an edge in a discrete pixel image. The SDSE and SDPE methods transfer an image of a marked region with scattered pixels into a continuous region which should be similar to the original marked region before loss of pigment particles and growth of fish. While a significant improvement has been achieved in the automated spray marking machine, and a satisfactory system has been developed for mark retention studies, further work could further improve the performance of the overall system. Several issues that may be addressed in future work are outlined below. 4.4 Emulsion Concentration The pigment to water ratio for pigment emulsion is crucial for the marking quality. In the beginning of a test, water and pigment are added to the mixer of the machine at a ratio that was experimentally found to give the best results. During operation, however, the used emulsion is recycled from the bottom of the machine and reused in the mixer. Because the recycled emulsion has been diluted with water carried by fish, and furthermore, fish carry away a part of pigment in the emulsion, a method to measure the emulsion concentration and to adjust the emulsion concentration automatically has to be developed in the future for maintaining the required pigment/water ratio. Chapter 4. Conclusions and Recommendations 65 4.5 Compressed Air The common source of compressed air for the machine is commercially compressed air cylinders. A full cylinder lasts for about 20 minutes of marking. Hence a large number of cylinders would be needed in the field site for extensive batch marking. It is inconve-nient to carry that many heavy cylinders to the marking site with the portable marking machine. An air compressor powered by gas engine that could continuously provide the required pressure in a regulated rnanner should be considered as an alternative. 4.6 Spraying Aerodynamics A marking region of a spray gun is schematically illustrated in Figure 2.3. For a given supply pressure of the spray system, the speed distribution of pigment grit inside the spray cone has to be estimated. The relationship between the percentage of the pigment grit which embeds into the fish skin for a reliable mark and the pigment grit speed has to be established in order to understand the nature of the marking region, and to control the mark quality. Because the mass densities inside and outside the spray cone are not the same, and also the density within the cone itself is not uniform, the speed distribution cannot be calculated easily, for exannple based on the aerodynamics of a pure air jet. 4.7 Marking of a C o d e Marking of a complete code rather than a simple presence/absence type of binary mark should be investigated next as a future objective in design of the automated spray marking machine. Two types of code: color codes and pattern codes, could be investigated. Obviously, more information on fish life could be marked and carried on the fish body if spray marking of codes are available. Chapter 4. Conclusions and Recommendations 66 Marking of codes would require proper positioning and orientation of fish. A zig-zag channel has been studied with the conveyor belt to constrain fish orientation. Based on tests, however, random fish jumping makes the design very difficult. 4.8 P igment Mark Retent ion Field testing of the newly designed prototype was done twice, in April and June of 1993. Since a complete fish life cycle is about three years, the period of pigment mark retention on fish body should preferably cover this duration. Some preliminary studies have been done in pigment mark retention, as described in Chapter 2, with pigment emulsion. It is clear that further investigation is needed over a three year period. 4.9 P igment Mark Detect ion Based on the pigment mark detecting technique developed in this work, as described in Chapter 3, more research is needed on how to filter out image noise due to condensed moisture on the camera lens and the water droplets on the fish body. The time offset between the red channel image and the green channel image has to be eliminated. With future development of a mechanical gear to place fish on the test platform and remove them automatically at the end of a test, the system for mark retention studies could be further improved. Appendix A Experimental Data on Zig-zag Shaped Feeder Channel This appendix gives the experimental data on fish orientation as they pass through the zig-zag feeder channel. These data were used in the statistical studies discussed in Chapter 2. There are eight groups of data from the experiment, as given in the following pages. Every two are based on the same experimental conditions. However, one is from the reading at Mi in Stage 1 and another from the reading at M2 in Stage 2 (Figure 2.11) of the feeder channel. As shown above, Z lMl , Z1M2, Z2M1, ... are the names of the data groups. The first letter \"Z\" stands for the zig-zag shaped channel; the second numeral denotes the different experimental conditions as given in Table A.4;and the last two characters indicate the location of the reading (i.e., either Mi or M2). The mean values of the three dimensions (mm) of the fish used in the test are given Table A.4: Experimental Conditions TESTS ZlMl Z2M1 Z3M1 Z4M1 Z1M2 Z2M2 Z3M2 Z4M2 EXPERIEMNTAL CONDITIONS S{TnTn) 63 38 38 76 45° 45° 60° 45° Ci('mm) 21 21 26 21 C2{mTn) 19 19 19 19 67 Appendix A. Experimental Data on Zig-zag Shaped Feeder Channel 68 in Table A.5; Table A.5: LENGTH WIDTH HEIGHT Dimensions of the Test Fish MEAN 122.2 13.7 27.5 STANDARD DEV 3.1 0.95 1.3 The orientation data (in degrees) from tests is given in eight pages from the next. Appendix A. Experimental Data on Zig-zag Shaped Feeder Channel 69 ZlMl: 0 -20 -30 -10 50 1 0 - 2 0 0 0 0 0 10 40 -70 -10 -60 0 40 30 60 -20 50 20 90 -20 -20 30 50 10 -10 10 -40 0 30 -50 -70 90 0 0 30 0 -40 50 30 20 40 0 0 40 40 -20 -30 -10 90 -40 -30 -10 30 -20 0 0 -10 20 0 20 0 0 40 -50 0 -10 -90 40 10 70 -30 -90 30 0 30 50 40 -40 -50 60 30 -60 20 0 70 40 0 40 70 0 -10 30 30 -20 40 -30 -40 0 40 -50 -20 50 0 0 -30 -70 -30 0 20 0 30 0 30 -20 40 90 0 -20 50 -30 90 70 80 0 50 0 20 0 0 -20 -30 30 -10 -80 0 50 30 90 -10 -40 -20 30 -20 -30 70 0 -20 -40 -10 -20 40 70 60 0 40 0 30 50 -40 40 0 50 10 20 -20 0 -30 -10 0 -30 30 0 80 60 10 70 20 0 -30 -30 40 -40 90 -70 20 20 -30 -40 70 0 -70 10 40 -30 -40 -10 -10 20 0 0 60 -20 30 20 0 30 50 30 30 20 -20 -20 80 20 30 70 10 0 20 -30 0 60 -10 -50 -30 0 10 50 40 -30 -90 30 50 -40 -70 0 -50 40 0 -20 0 40 70 0 20 -50 -20 -50 30 -30 40 0 40 0 0 0 -40 Appendix A. Experimental Data on Zig-zag Shaped Feeder Channel 70 Z1M2: -20 0 10 0 -30 -20 -10 -10 -30 0 0 50 0 0 0 30 60 -20 0 40 20 -10 -20 80 40 0 60 10 20 -40 -10 20 0 0 90 0 20 70 0 10 20 -10 0 -30 10 30 10 -10 30 30 40 20 -20 -20 0 0 -30 -20 50 10 -10 -20 0 -20 -10 0 10 -10 0 0 -90 -60 30 -30 -10 -10 -10 20 0 60 -20 -30 -10 90 -40 -30 -10 30 -20 0 0 -10 20 0 20 0 0 40 -50 0 -10 -90 40 10 70 -30 90 30 0 30 20 20 20 0 0 -10 0 0 40 0 0 0 50 0-20 0 -10 10 30 0 -40 -10 0 -10 -30 60 70 20 0 0 20 10 20 0 0 10 0 -40 0 0 -50 -50 -70 -20 -30 0 0 0 -30 0 10 20 30 -40 30 40 90 -20 -40 70 0 0 0 90 -20 30 10 -30 0 -20 0 10 -10 0 0 -10 10 10 10 -20 0 0 0 20 0 -20 20 20 0 30 0 0 -80 0 -10 0 -10 30 0 -30 -20 -30 50 0 -10 0 40 60 -10 20 10 40 20 0 0 10 0 -20 0 -40 -20 0 40 0 0 0 -30 -20 0 -30 0 0 0-10 Appendix A. Experimental Data on Zig-zag Shaped Feeder Channel 71 Z2M1: 20 30 -30 -40 0 0 0 0 -40 10 -40 20 0 -40 10 -20 -30 30 0 -20 -50 0 -30 -20 0 -30 70 0 0 0 -70 0 20 70 -10 0 30 10 -70 80 0 -70 90 10 10 -50 -40 -40 40 30 -20 60 0 20 30 60 -30 0 -30 -70 40 0 -90 -30 30 10 70 30 -40 40 20 -10 30 20 30 -20 0 10 20 0 50 -20 0 20 -30 70 0 0 -20 -30 -50 20 0 -10 10 30 20 30 60 30 30 -50 0 90 -10 -10 20 0 0 0 -40 20 0 -10 10 -10 -20 20 70 80 60 80 -20 20 30 0 -20 -10 30 40 -10 -10 -40 0 -10 20 -10 -40 0 20 30 20 -10 0 30 -30 -80 20 40 -70 60 -20 0 - 2 0 0 0 40 -10 Appendix A. Experimental Data on Zig-zag Shaped Feeder Channel 72 Z2M2: 0 -10 -20 0 0 0 0 0 -10 0 0 -20 -10 30 20 0 -20 0 30 40 60 0 20 20 -50 30 0 80 20 0 10 10 -30 0 10 30 20 50 0 0 20 10 50 10 0 0 0 0 10 10 0 0 0 0 0 0 0 80 50 60 0 0 0 0 0 0 0 0 10 50 -40 -10 0 -70 -70 0 0 70 -30 -20 60 -70 0 - 9 0 0 0 -10 0 0 -10 0 0 -20 -10 0 0 20 0 - 1 0 0 -10 -10 -50 0 0 20 50 -10 40 0 0 90 0 40 -80 0 0 0 -40 0 0 - 2 0 0 0 0 - 5 0 0 0 0 0 10 10 20 0 -90 -10 0 60 -30 -40 90 80 0 40 0 0 0 -30 -10 -20 20 10 0 40 0 40 20 -70 30 0 10 -60 -20 0 -70 -20 50 -80 0 0 30 40 0 0 0 -40 20 -20 40 0 -30 0 0 50 20 10 0 0 0 0 0 0 - 1 0 50 10 0 0 0 30 -30 -40 0 0 Appendix A. Experimental Data on Zig-zag Shaped Feeder Channel 73 Z3M1: 40 -50 60 0 0 -30 -70 30 -10 20 20 0 -40 -70 30 -80 -40 70 0 30 -40 -20 -40 10 20 20 0 10 -30 -20 30 -60 -70 0 40 30 -30 -20 20 80 -40 -60 10 20 0 50 -30 0 0 -20 40 -50 0 40 -40 -20 -20 0 20 20 -40 0 -20 40 0 0 -30 20 -80 70 90 -40 -20 -30 20 40 0 20 30 10 -50 20 -70 -80 -80 0 40 -60 -30 30 30 -50 0 -70 -50 30 -40 30 -20 -20 40 0 0 50 -40 -80 -70 30 0 40 -70 90 -40 40 -20 30 0 -30 0 40 70 -20 -20 -10 -30 30 -40 -90 40 30 30 40 40 60 70 80 0 10 50 30 20 -20 0 40 -30 0 -20 90 50 50 -30 20 -20 -30 -20 0 0 -30 -10 50 30 -10 -20 50 -60 80 -30 -20 -10 -30 -40 -20 -90 40 50 -30 -40 80 -30 -30 20 0 30 40 -60 0 -40 40 -30 0 -50 40 -40 0 20 0 0 -40 10 70 20 20 20 30 60 90 -50 40 30 -70 40 50 30 30 0 -10 10 40 -80 -90 -70 Appendix A. Experinnental Data on Zig-zag Shaped Feeder Channel 74 Z3M2: 10 20 -20 -40 -40 20 0 -50 -30 -80 40 10 0 -20 0 40 0 -30 90 -80 -30 30 20 0 -40 0 -10 -10 0 10 0 -80 80 -20 -60 -10 0 70 -50 0 0 10 30 70 -20 -10 60 -30 -30 -30 20 0 0 -20 0 -70 -60 80 -60 -70 70 0 - 6 0 0 0 0 -30 -40 -20 50 -80 -20 70 10 20 -30 -20 -10 -10 10 -10 -20 -10 10 -30 40 -10 0 -30 -20 20 10 90 0 -10 -90 0 0 10 0 10 20 -70 0 -80 10 -40 40 60 50 -20 -10 -20 -20 -50 0 40 30 -60 -70 -50 -60 -40 -20 0 0 -80 -10 0 -50 0 -50 70 50 0 0 0 - 7 0 0 90 -70 30 -20 20 -60 40 80 10 -10 40 70 -30 -60 0 10 20 -80 20 0 -30 -40 -50 0 0 0 10 20 -30 0 -30 20 30 70 80 0 -10 0 -10 70 -20 -20 20 0 -20 0 -70 -20 0 0 -30 20 10 0 0 0 -10 -60 -20 10 20 70 -90 Appendix A. Experimental Data on Zig-zag Shaped Feeder Channel 75 Z4M1: 0 -20 -10 10 0 40 50 -70 0 -60 10 -30 10 10 30 0 -10 -30 60 10 -50 30 60 0 -60 40 -80 30 0 0 20 -20 0 -40 0 -10 0 -20 -20 -40 50 0 -10 20 50 0 40 80 10 -20 -40 -30 20 0 40 -40 -30 10 -10 20 80 0 0 -30 -40 0 60 40 0 70 20 10 -70 0 0 -60 40 90 20 40 10 50 40 50 60 60 70 -20 -40 20 10 -90 -40 0 0 70 80 60 -70 -70 0 50 10 -70 20 30 40 30 30 -40 0 -30 50 30 50 10 40 0 -40 10 0 0 -80 90 -90 60 -10 70 10 0 60 -30 0 -30 0 0 0 30 30 -20 0 0 10 80 70 -70 0 -80 -30 0 50 90 -30 -20 -30 -20 10 30 -30 20 -30 0 40 70 0 -30 40 20 10 -10 10 -10 -30 30 -70 0 0-50 0 -20 Appendix A. Experimental Data on Zig-zag Shaped Feeder Channel 76 Z4M2: -20 -10 -40 -20 0 70 0 0 -30 -20 -30 -40 -10 10 70 40 80 -70 50 -30 -10 20 20 -40 -10 -60 -80 70 60 -70 -50 0 -10 -10 10 -30 30 50 60 0 -10 -20 -20 -10 0 0 0 -50 10 40 70 0 20 0 20 0 -20 60 0 -20 0 0 20 -20 70 30 30 0 0 0 20 50 0 -40 10 0 -10 -20 0 70 -10 0 60 -50 -40 -50 -10 -30 40 10 -10 10 0 70 -80 20 0 0 -30 30 20 0 -10 80 30 -20 -60 -10 -20 -70 -20 -20 -30 0 -50 -10 -20 0 30 -40 0 -20 80 20 -60 -10 90 50 -10 -30 -20 0 0 40 10 0 0 -30 -20 10 10 10 0 -10 0 - 2 0 0 0 -20 -10 0 20 0 0 70 30 -70 0 -70 0 10 -70 0 20 -10 -10 -10 -70 0 10 10 20 40 -10 -20 30 10 0 0 -10 0 10 50 -10 -10 -20 30 10 20 -10 0 -20 30 -20 -40 -60 -10 0 80 -90 60 -20 0 -40 30 20 0 -20 20 10 -10 -30 -30 50 -40 -30 0 20 -20 10 20 0 -30 Appendix B S D S E M e t h o d of Connection between Pixels (k — 2) This appendix presents several example cases of image processing results using the SDSE method that was described in Chapter 3. The diagram for the cases is given in the next two pages. 77 Appendix B. SDSE Method of Connection between Pixels (k — 2) 78 Case No. The Original Image The Final Image The increment of Pixels 0 iJm • • • • • • I ^m ••M • • • • ym «•• >••! K » • ' • • • • • • • • • I w n K 8 i i. (Con't next page) Appendix B. SDSE Method of Connection between Pixels (k = 2) 79 Case The Original No. Image The Final Image The increment of Pixels 10 TTT r m r r r n m TTT- T T T * ' ' iXn 111 T T T t\"l I TTT T T T i X P T11 i IT Ijiiliii iiiijul. ^^T ^ ^Q ^ ^T ^ ^3 T^E ^ ^ i TTlT111 ITT f f 1 1 l l TTT l i « i h » . i . « i » . » J laBt • • • IMMU mm tmwt wmw rwn P •••*^--=*=—* — ^ ^•' 32 11 27 12 23 13 I mmm i »»] m Appendix C S D P E Method of Connection between Pixels {k = 2) This appendix presents several example cases of image processing results using the SDPE method that was described in Chapter 3. The diagram for the cases is given in the next four pages. 80 Appendix C. SDPE Method of Connection between Pixels (k = 2) 81 Case The Original No. Image The Final Image The increment of Pixels m m \"*\"*\"l!a 8 m • \" i —Spw M 10 ^Wv miHis (Cont'd next page) Appendix C. SDPE Method of Connection between Pixels (k = 2) 82 Case No. The Original Image The Final Image The increment of Pixels 11 • • • ! • • • : ^ S 12 • • • laBi •aa mmmSSi j a . . a g g a 13 ^ 14 15 W • • • I BS • • • ! 16 ssus: • • • : Si 17 ^ siu »ai aai m (Cont'd next page) Appendix C. SDPE Method of Connection between Pixels (k = 2) 83 Case The Original No. Image The Final Image The increment of Pixels 18 19 0 20 0 21 (Cont'd next page) Appendix C. SDPE Method of Connection between Pixels (k — 2) 84 Case No. The Original Image The Final Image The increment of Pixels 22 rwn III HMIm' ! • • ! • • • ! • • • • • • m^. 13 23 :!!I.!SH|::: 14 24 12 Bibliography [1] Jackson, C. P., \"A technique for mass-marking fish by means of compressed air\", New Hampshire Fish Game Dep. Tech. Circ, Vol.17, ppl-8, 1959. [2] Phinney, D. E., Miller, D. M., and Dahlberg, M. L., \"Mass marking young salmonids with fluorescent pigment\", Transactions of the Am.erican Fisheries Society., Vol.96, ppl57-162, 1967. [3] Hennick, D. P., and Tyler, R. W., \"Experimental marking of emergent pink salmon fry with sprayed fluorescent pigment\". Transactions of the American Fisheries So-ciety, Vol.99, pp397-400, 1970. [4] Phinney, D. E., \"Growth and survival of fluorescent marked and fin clipped salmon\", Journal of Wildlife Management, Vol.38, ppl32-137, 1974. [5] Ware, F. J., \"Mass marking warm water fish by compressed air and fluorescent pigment\", Proceeding of the Annual Conference Southeastern Association of Game and fish Commissioners, Vol.22, pp339-342, 1969. [6] Pierson, J. M., and Bayne, D. R., \"Long-term retention of fluorescent pigment by four fishes used in warm water culture\" Progressive Fish-Culturist, Vol.45, ppl86-188, 1983. [7] Moodie, G. E. E., and Salfert, I. G., \" Evaluation of fluorescent pigment for marking a scaleless fish, the Brook Stickleback\", Prog. Fish Cult., Vol.44, ppl92-195, 1982. [8] McAfee, M. E., and Loucks, G. A., \"An air powered device for dispensing fluorescent marking pigment\". Prog. Fish Cult., Vol.48, pp72-74, 1986. [9] Pauley, G. B., and Troutt, D. A., \"Comparison of three methods of fluorescent dye application for marking juvenile steelhead\". Trans Am. Fish Soc, Vol.117, pp311-313, 1988. [10] Whitmus, C. J., \"The influence of size on the early marine migration and mortality of juvenile chum salmon (Oncorhynchus keta)\", A thesis submitted on Master of Science in University of Washington, 1985. 85 Bibliography 86 [11] Dong, C , de silva, C. W., and Gosine, R., \"Particle Retention Measurement using Image Processing for a Spray Marking System\", Proc. IEEE Pacific Rim, Conference on Communications, Computers, and Signal Processing, Victoria, BC, Vol.2, pp.790-793, 1993. [12] Horn, B. K. P., \"Robot vision\", Cambridge, Mass.: MIT Press, 1986. "@en ; edm:hasType "Thesis/Dissertation"@en ; vivo:dateIssued "1994-05"@en ; edm:isShownAt "10.14288/1.0080825"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Mechanical Engineering"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en ; ns0:scholarLevel "Graduate"@en ; dcterms:title "Development of an automated marking machine for live fish with application in fishery management"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/4971"@en .