UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Some television bandwidth-compression systems using edge coding Farr, Peter 1966

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1966_A7 F3.pdf [ 7.5MB ]
Metadata
JSON: 831-1.0103228.json
JSON-LD: 831-1.0103228-ld.json
RDF/XML (Pretty): 831-1.0103228-rdf.xml
RDF/JSON: 831-1.0103228-rdf.json
Turtle: 831-1.0103228-turtle.txt
N-Triples: 831-1.0103228-rdf-ntriples.txt
Original Record: 831-1.0103228-source.json
Full Text
831-1.0103228-fulltext.txt
Citation
831-1.0103228.ris

Full Text

SOME TELEyiSION^BANDWIIJTH-COMPRESSION SYSTEMS USING EDGE CODING JOHN PETER PARE B.E.(Hons 0) r University of Western A u s t r a l i a , 1963 A THESIS SUBMITTED IN PARTIAL PUIPILMENT'-OP THE REQUIREMENTS POR THE DEGREE OP MASTER OP APPLIED SCIENCE i n the Department of E l e o t r i c a l Engineering We accept t h i s thesis as conforming to the required standard THE UNIVERSITY OP BRITISH COLUMBIA APHIL, 1966 p In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t of the r e q u i r e m e n t s f o r an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree t h a t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and st u d y , I f u r t h e r agree t h a t p e r -m i s s i o n f o r e x t e n s i v e c o p y i n g of t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the Head of my Department or by h i s r e p r e s e n t a t i v e s . I t i s understood t h a t , c o p y i n g or p u b l i -c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n . Department of ^1-*. c-Anctf gje^,,^ The U n i v e r s i t y of B r i t i s h Columbia, Vancouver 8, Canada Date ABSTRACT The results of a f e a s i b i l i t y study of soifie""hew~-" methods fo r reducing the bandwidth required for transmission of commercial-quality t e l e v i s i o n are described i n t h i s thesis. It i s shown- that t e l e v i s i o n requires a wide transmission bandwidth because of the broad frequency spectrum obtained i n scanning scenes which contain abrupt, changes of. brightness at edges* An e f f i c i e n t method of coding these edges i s therefore sought. The properties of the Gabor-type hologram are studied and i t i s shown that holograms may be used advantageously to code the picture edge information. A two-channel system i s proposed ' and i t s construction and t e s t i n g described* The low-frequency information i s transmitted i n conventional form on one channel and the high-frequency information i n hologram form on the other channel. The test results support the theoretical prediction that the system can transmit t e l e v i s i o n pictures with reduced bandwidth* However? the system i s shown to have p r a c t i c a l disadvantages lar g e l y caused by the necessity of making electro-o p t i c a l conversionswhich cannot be performed i n r e a l time. In one of the alternative systems proposed here 7 a reduced bandwidth i s made possible by transmitting the low-frequency information conventionally ? and the edge pattern i s scanned i n a - v e r t i c a l d i r e c t i o n to transmit the high frequencieso In a p r a c t i c a l test, the system gave a bandwidth compression r a t i o of 6:1, n-As a basis for another possible system i t is shown that because the edge pattern of typical pictures occupies only a small fraction of the raster area and exhibits a high degree 1 of spatial correlation, bandwidth compression could be achieved by transmitting only this edge information. i i i TABLE OF CONTENTS Page LIST OF ILLUSTRATIONS . . » » . . , . o . « . o o » « . o . . . . . . . « o o o o o o v i ACKNOWLEDGEMENT . o . o o . o o o o o o . o o o e o o . o o o o . o c . o o o D . o o o o o v i i i 1o INTRODUCTION 9 O . . o . o o o . o o o 0 . o . o o T > a o o o o o . o . o . o . o . o o 1 2. THEORETICAL BASIS FOR A COMPRESSION SYSTEM USING HOIfOGrR-AjyiS O O O O O O O O O O O O O O O D O O O O O O O D O O O O O O O O O O O O O O O O 5 2.1 Analysis of the Conventional Television Frequency Spectrum for a C r i t i c a l Case ...... 6 2.2 Theory of the Gahor Hologram Process ........ 8 2.3 Analysis of the Hologram Frequency Spectrum . 12 2*3.1 Theoretical Analysis 13 2.3.2 Experimental "Verification ............ 16 2.4 Description of the Two-Channel System 18 3. ELECTRONIC INSTRUMENTATION ....................... 22 3 o 1 IidW 0 , "l?9 .SS Fll"b©I*S oaoatioaxia»oom»oooo»t>ooQoeoo9 22 3 e 2 Edge Detector • „ •» ..» 26 3.3 Reconstruction F i l t e r . . . . . . . . o . o . o o . . . . . . . . . 30 3.4 Tests of the Electronic Instrumentation ..... 36 4. OPTICAL INSTRUMENTATION »..•.............•.....„.'.. 38 4.1 Hologram Formation from Edge-Detector Display 38 4.2 Hologram Reconstruction Process . „ „.......... 41 4.3 Real-Time Recording of Holograms and Reconstructions 42 5. ALTERNATIVE BANDWIDTH-COMPRESSION SYSTEMS USING ED GrE~0 OD INGr O O O O O O O O O O O O O O O O O O O O O O O O O O D - O O O O O O O D O S S 43 5.1 Practical Disadvantages of the Hologram Transmission System 43 5.2 An Alternative Two-Channel System using Edge-5.3 Basis for a Proposed One-Channel System using Edge-Coding 47 iv \ Page E) • TEST RESULTS » » O 0 e o » O O 0 0 0 0 0 O 0 0 0 D 0 0 O O 0 0 0 0 0 0 0 0 0 O 0 0 0 O 0 0 0 49 6.1 The Two-Channel System using the Hologram Principle .. 49 6.2 The Alternative Two-Channel System using Edge-Coding 53 *Y » C 0N0 LUS IONS 0 O O 6 O 0 O D O O Q 0 O C t 0 0 O 0 0 O 0 O 6 0 0 O 0 0 0 » O 0 t r 0 O 9 0 0 0 » » 5^  REFERENCES • o o « o o o d o o o o o o © o o o o o o o o o o o o o o o o o o o o » o o o o o o » » « » V LIST OF ILLUSTRATIONS Figure Page 2-1 R e l a t i o n s h i p between Scanning Raster and Test 2-2 • Camera Output from Scanning P a t t e r n Shown i n Flj^HX*© 2 — 1 • 0 0 O O * 9 « * « « a 9 9 O « O O 0 O 9 S S t O S * O O S O O « O * * 7 2-3 Frequency Spectrum of Waveform Shown i n Figure 2-2' 7 2-4 I l l u s t r a t i n g Gabor's Method of Imaging by Reconstructed Wavefronts 9 2-5 Frequency Spectrum Obtained i n Scanning the Hologram of a Rectangular Aperture 14 2-6 V e r t i c a l Black Bar (a) and i t s Hologram (b) ... 17 2-7 Block Diagram of T e l e v i s i o n Bandwidth-Compression System 19 2- 8 Waveforms at Labelled Points of the System Shown i n Flc^liX*© 2"~ 8 • • • • o e 9 O 0 9 a 9 O 9 0 a a g a s 9 9 o o o o o o a 9 0 0 « 20 3 - 1 Transfer-Function Pole P o s i t i o n s f o r Third-Order FllijSHTS • • O O O 6 O 0 0 9 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O 0 O * 2^ 3-2 1.0 Mc/s Low-Pass F i l t e r ...................... 24 3-3 Frequency Response of 1.0 Mc/s Low-Pass F i l t e r . 25 3-4 Source f o r Low-Pass F i l t e r of Channel #1 . . . . . . 25 3-5 T r a n s i s t o r D i f f 0 X* 6H.~fc lEl"fc 0 X* 0 0 O 0 0 9 D O O O O O O O O O O 9 0 0 0 2^ 3-6 D i f f e r e n t i a t o r Waveforms ...................... 28 3-7 Modified D i f f e r e n t i a t o r ....................... 29 3-8 Block Diagram of System Omitting Hologram PX*0 C S S S • O 0 9 0 0 0 0 0 O 0 9 » 0 0 O 0 0 9 O O 0 * 0 O « C > O 0 0 9 0 0 O e * 0 0 O 30 3-9 Block Diagram of Reconstruction F i l t e r H(s) ... 31 3-10 Elementary D i f f e r e n t i a t o r ..................... 32 3-11 Modified D i f f e r e n t i a t OX* 0 0 9 0 0 a 0 0 0 0 9 6 0 0 0 O 0 9 0 0 0 0 0 32 3-12 Integrator Complementary to D i f f e r e n t i a t o r of Figure 3-11 33 v i Figure Page 3-14 Part of Reconstruction F i l t e r .. 35 3- 15 System Waveforms for a Pulse Input Signal 37 4- 1 Experimental Arrangement for Recording Holograms 39 4-2 Photograph of Hologram Recording Apparatus .... 40 4- 3 Experimental Arrangement for Obtaining Reconstruction 41 5- 1 Block Diagram of System using Edge-Coding 45 5- 2 Test Scene (a) and i t s Edge Pattern (b) 48 6- 1 Feasibility Study Results for Hologram System . 51 6-2 Test Results for Two-Channel Edge-Coding System 54 v i i ACKNOWLEDGEMENT Acknowledgement is due to many persons who have helped" during the course of this research project. In particulars, I -would like to thank Dri M0P<i Beddoes and Dr* A»D. Moore for guidance and encouragement while supervising the project? Dr. J.S« MacDonald and Mr* S^ A, Akhtar for helpful suggestions and discussions $ Mr» A, Horn for assistance with the photographic work? and Mr. L-. Irvine and Mr„ M. Bishop for their co-operation in making the television studio and associated equipment at the British Columbia Institute of Technology available for test purposes. The financial support received from the National Research Council through NEC Grant A-68 is also gratefully acknowledged. v i i i 1. INTRODUCTION An important factor which limits the scope of t e l e v i -sion as a communication medium is the relatively large bandwidth required* For instance, commercial television conforming to the North American standard of 30 frames per second, 525 lines per framej requires a bandwidth of about 5 Mc/s. In at least two particular instances, there are incentives for finding methods to reduce the bandwidth of the transmitted information without impairing the subjectively estimated quality of the received pic-ture . These cases are in television relay links between large population centres, and in the "television phone". At present, coaxial cables, microwave relay systems and active a r t i f i c i a l satellites are used on a limited basis to pro-vide television relay links between some of the cities of the world. A reduction in television bandwidth would reduce the complexity and cost of these transmission channels or allow more programs to be broadcast simultaneously on the.existing channels, and allow such existing limited-bandwidth channels as trans-oceanic cables to be used for television links» Again, the television phone w i l l not be possible on a commercial basis until a system giving considerable bandwidth compression is developed. -Present-day television systems transmit information about every picture element. The bandwidth required depends upon the line and frame scanning rates and the number of elements (approximately 600) per line. If i t were s t a t i s t i c a l l y true that 2 each picture element was of independent brightness, the bandwidth of the present system could not be reduced. It has been shown in a number of investigations that the transmission channel capacity requirements may be substantially reduced. 1 This possibility arises from three separate and distinct forms of s t a t i s t i c a l redundancy. Fi r s t l y , the brightness variations along any given scanning line in the picture are not random; relatively large areas of any scene are of constant or near constant brightness, so that correlation exists between successive picture elements. Secondly, i t i s probable that adjacent scanning lines w i l l contain identical or nearly identical information. Thirdly, successive pictures are identical unless movement has occurred, and even then w i l l differ only in those regions where movement has taken place. Methods can be devised, at least in theory, which w i l l remove any or a l l of these redundancies in the transmitted informa-tion, thereby reducing the necessary channel capacity. It i s important to note, however, that mere removal of redundancy w i l l not bring about a reduction in signal bandwidth, but only in signal power. In order to achieve transmission in a reduced bandwidth, i t i s necessary to redistribute the non-redundant information on the time axis, so that the information is transmitted at a more uniform rate. This requirement was f i r s t described for a general p transmission channel in the classical paper by Shannon. There exists a fourth, but less well known, avenue for bandwidth reduction — that of u t i l i z i n g certain properties of the human sense of vision. Seyler and others have shown that in the 3 important cases of (a) picture areas containing fine detail, (To) complete changes in the scene "being televised, (c) panning "by the camera and (d) moving objects, the number of contrast levels which can be discerned is substantially reduced. Thus, man's abi l i t y to resolve spatial detail deteriorates in these instances and this permits a reduction in the rate of data transmission. It is apparent that the optimum compression system w i l l use both s t a t i s t i c a l redundancies and the psyc'hophysics of vision. Some experimental systems incorporating one or more of the above design philosophies have been tested, 1'but because of the complexity of the processing equipment and the small compression ratios achievedf> no system, (excepting simple low^pass filtering) has been adopted on a commercial basis. In recent years much work in this f i e l d has been done by d i g i t a l computer simulation. In a two-channel scheme simulated 4 at M^I . To , compression ratios of about 10si have been reported. The received pictures had rather poor quality even though these results were obtained under highly idealized conditions, especially with regard to spatial linearity in the two channels and the presence of noise. The original objective of the author's research was to conduct a f e a s i b i l i t y study of a bandwidth-compression system having the unique feature of transmitting high-frequency video information in hologram form? a technique developed by S.A. Akhtar and the author. Since a l l parts of the system could not be instrumented for real-time operation, s t i l l - t e s t slides had to he used in the fe a s i b i l i t y study. However, the extra instrumen-tation required for real-time operation is also described. During the course of the fe a s i b i l i t y study i t was found that the hologram transmission system had some practical disadvantages, and two related, but simpler, systems were devised to overcome these d i f f i c u l t i e s . A preliminary f e a s i b i l i t y study of these two systems is described. 5 2 . THEORETICAL BASIS FOR A COMPRESSION SYSTEM USING HOLOGRAMS Typical television scenes contain areas of slowly varying "brightness separated by relatively sharp discontinuities at the edges. Low frequencies of the television signal are produced by gradual variations of spatial brightness, while edges not oriented in the direction of the scanning lines give rise to h£gh frequencies. A reduction of transmission channel bandwidth w i l l therefore be possible i f these two basic classes of signals can be coded efficiently. If a video signal is reduced in bandwidth by a factor of about 10 by low-pass f i l t e r i n g and the resultant signal applied to a television monitor, those picture edges not oriented in the direction of scanning w i l l be blurred. Otherwise the picture 'is unaffected? edges oriented in the scanning direction a are clearly defined and correct contrast is maintained. This indicates that the conventional television system is i t s e l f an efficient transmitter of the low-frequency information. Consider a video signal containing frequencies in the range 0 ^ w ^ Wq and assume that the low-frequency portion, 0 ( u { aco0, where a < 1 . 0 , i s transmitted on one channel. It is shown in this chapter that the complementary spectrum acoj < oo ^ w 0 may be transmitted in hologram form on a second channel in a bandwidth less than (w^ - awQ). Using this fact, a two-channel bandwidth-compression system was devised and is described in Section 2 . 4 . In this system, the low-frequency components of the video signal are transmitted in conventional form on one channel and the high frequencies in hologram.form 6 on the other channel. At the receiver the hologram signal is processed to recover the high-frequency information and this is then superimposed with the low-frequency signal. The unique aspect of the system, relative to other systems using separate transmission of the low- and high-frequency information, is the conoept that "bandwidth-compression may be obtained by converting the high-frequencies into hologram form for transmission purposes. 2.1 Analysis of the Conventional Television Frequency Spectrum  for a C r i t i c a l Case A simple example of the manner in which high frequen-cies may be produced in a conventional video signal is that of sequential scanning in the horizontal direction of a thin black bar oriented in the vertical direction, the background being white (Figure 2-1). Figure 2-1 Relationship between Scanning Raster and Test Pattern — — — — — — — / 7 The signal current from the television camera w i l l "be a pulse train of the form shown in Pigure 2-2. In this figure, for a spot velocity v, d = a/v and T = W/vD Camera Output A \ — ( — — i — 3» -T d n d T Time ~2 U 2 Figure 2-2 Camera Output from Scanning Pattern Shown in Pigure 2-1 The pulse train gives' rise to a line frequency spectrum, the envelope of which has the equation q/ M l Ad . sin tod/2 l o , \ S ( w ) = ~ ud/2 ' ( 2 ' 1 } This i s plotted for positive frequencies in Pigure 2-3. S(to) d d d d d Pigure 2-3 Frequency Spectrum of Waveform Shown in Pigure 2-2 8 This function lias an infinite "bandwidth "but the amplitude of the function tends to zero as the frequency tends to i n f i n i t y . To give an exact reproduction of the input signal, a l l components of the conventional television system should he capable of transmitting this infinite range of frequencies,, In practice, the bandwidth of the system is restricted to about 5 Mc/s so that an exact reproduction of such an input signal is not possible. However, due to the psychophysics of vision^ the quality of such a band-limited reproduction is acceptable to most viewers. Pigure 2-3 also shows that for a rectangular pulse train, the proportion of the total frequency spectrum contained within a given low-frequency range is proportional to l/d, and is therefore inversely proportional to the pulse duration* Thus, the bandwidth required to transmit a vertical bar with conven-tional television is inversely proportional to the width of the bar. Therefore, i t is edges of brightness in the input scene which give rise to the high frequencies of a video signal. It is shown in the next sections that the edges may be transmitted within a relatively small bandwidth in hologram form. 2.2 Theory of the G-abor Hologram Process Consider a coherent wave from a source S, impinging on a semi-transparent object 0 (Pigure 2-4). Let H be a photo-graphic plate some distance behind the object and let U = A e i a represent the complex disturbance at a typical point P on H, A 9 Coherent •light Source Object t(x.y) y H Hologram to y CO X (a) Formation of the Hologram Positive Hologram (h) Reconstruction Figure 2-4. Illustrating Gabor's Method of Imaging by Reconstructed Wavefronts being the real amplitude, and a the phase of the disturbance. U may be regarded as the sum of two terms, U = U + TJ = A e i ao + A, e i a l o 1 o 1 = e o A + A i e i ( a l " ao) . o 1 (2.2) 10 Here U q = AQe 0 denotes the background, or carrier, wave and is the f ie ld which would be produced at H in the absence of the object. The other term, = A-^ e -^represents a secondary, or diffracted, wave and contains information about the object. Prom ( 2 . 2 ) , the * * r DTJ L A 2 + A 2 + 2 A Q A 1 cos(a 1 - aQ) J „ ( 2 . 3 ) It can be shown that a positive print of the photographic record ( i . e . hologram) at H, has transmission function • t = KA (2.4) where, for given films developed under standard conditions, K and t are constants dependent upon the product of the l ight intensities and exposure times used in the positive and negative processes. If, in the reconstruction process, the positive + \ •' hologram (H ), whose amplitude transmission factor t is given by Equation ( 2 . 4 ) , i s illuminated by the coherent background wave alone, a 'substitute wave', TJ , w i l l be transmitted by the hologram plate, and according to Equations ( 2 . 3 ) and ( 2 . 4 ) ? this is represented by U = TJ t* = KA e i a ° S O 0 0 0 I An + A^ + 2A A, cos (a, - a ) O 1 0 1 i - 0 (2.5) As can be shown, the simplest and most advantageous choice for <f is (f = 2 which gives 11 TJ = KA e i ao [A? + A? +-2A An cos (a, - a ) 8 o L o 1 o l 1 0' = KA 2e i ao o A? (AQ + A ^ 1 ^ ! " ao)) + + A ^ " 1 ^ ! " "o)) (2.6) On comparing Equation (2.6) with Equation (2.2) i t i s seen that i f A Q is constant,, i.e. i f the "background wave i s uniform, the substitute wave TJg contains a component, called the reconstructed wave, proportional to the original wave TJ (the f i r s t and second terms of Equation (2.6)). The remainder of Equation (2.6) consists of two terms. One has the same phase as the background wave and an amplitude A^/AQ times that of the background. This term can be made small, i f the background i s sufficiently strong. The other term has the same amplitude as- the reconstructed wave, but has a phase shift of opposite sign relative to the background, and is called the conjugate wave. The conjugate wave may be regarded as being due to a fi c t i t i o u s object similar to the true object, but situated in a different plane.- In practice, the conjugate wave of the Gabor process provides an out-of-focus background image to the required reconstructed image. These unwanted effects can be removed but not easily. In recording the 'split-beam' type hologram developed by Leith and Upatnieks,^ the carrier wave impinges on the photo-graphic plate at an-angle to the optical axis,- and in the re-construction this causes the conjugate image to be formed in the same plane as the required reconstructed image, but angularly separated from this image. High-quality reconstructions may be 12 obtained using this method for two^tone, continuous-tone and three-dimensional objects. However, because the carrier wave reaches the photographic plate at an angle to the diffracted wave, the interference fringes are very closely spaced, and i t has been shown that a very wide bandwidth would be required to 7 transmit such a hologram by television. Consequently, the split-beam method cannot be used in the television bandwidth-compression system being investigated in this thesis. It is shown in Section 2.4 that, in the fea s i b i l i t y study, Gabor holograms w i l l be formed of objects consisting of arrays of black and white bars (corresponding to brightness edges in the continuous-tone pictures) on grey backgrounds, recorded in the form of photographic transparencies. The carrier wave can reach the hologram plane by transmission through the grey background, and consequently the theory developed above applies. The hologram reconstructions should be of quality comparable to that demonstrated by G-abor for a similar class of objects consisting of black bars on transparent back-grounds.^ 2.3 Analysis of the Hologram Frequency Spectrum In order to obtain an estimate of television bandwidth requirements for transmission of Gabor-type hologram= this section deals with the hologram frequency spectrum from theoreti-cal and experimental considerations. Since we are concerned with forming the holograms of edge patterns consisting of arrays of narrow bars, the special case of a single vertical 13 bar w i l l be considered„ 2,3»1 Theoretical Analysis Consider a two-dimensional object with non-negative optical transmission function t(x,y) = t Q(x,y) + ±^(x,j)f where t is constant and t n i s variable. When the object is o. 1 0 illuminated with a coherent monochromatic light wave, t gives rise to the carrier wave U as discussed in Section 2,2* and o v similarly, t-^  gives rise to the diffracted wave (Pigure 2-4)« Neglecting the uniform wave transmitted by t Q , i t can be shown that the Praunhofer diffraction pattern at a point P in the to to spatial frequency plane due to this object is given by x y the Pourier transform equation - j (to x + to y) U(P) = U(a>x,wy) = C I/ t 1(x,y)e x ? dxdy, (2,7) where C is a constant, a is the object plane, and to and to x y represent spatial frequencies (radians per unit length) in the g x and y directions, Por the special case of a long rectangular aperture of width 2a in the x direction, Equation (2,7) simplifies to , sin aw U(P) =U(wz) = C , (2,8) x i where C is a constant. It has been shown by Akhtar that the time function obtained i n scanning the spatial amplitude pattern given by Equation (2,8) with a spot moving in the x direction at a 14 velocity v is 0(t) = C , sin to^t (2.9) where to^ 2«av/A.s , A. is the wavelength of the monochromatic light source producing the hologram, and s is the distance q from the aperture to i t s hologram. It i s easily shown that 0(t) is the inverse Fourier transform of the frequency function, E(io) = C /2u>^ for -to-^  < to < = 0 for 1 | > to. (2.10) Hence the frequency spectrum of the!electrical signal obtained in scanning the hologram of an aperture of width 2a is as shown in Figure 2-5. B(OJ) C 2ton to^ Frequency-to Figure 2-5 Frequency Spectrum- Obtained in Scanning the Hologram of a Rectangular Aperture Therefore? the bandwidth of the spectrum obtained in scanning the hologram is 15 It i s seen from Equation (2.11) that as the aperture width 2a decreases the bandwidth obtained in scanning the hologram decreases. By contrast, i t was shown in Section 2.1 that, for a conventional television system, bandwidth requirements increase as the bar width a decreases. It is now possible using Equation (2.11) to estimate the order of magnitude of the bandwidth required for hologram transmission. Consider an object obtained by photographing the television monitor display of an edge pattern consisting of an array of vertical bars. Assume that the monitor displays 600 picture elements per line and that the width of the object (i.e. the-monitor photograph) is 2,5 cm. It i s reasonable to assume that the maximum width (2a) of any bar in the edge pattern w i l l be 4 picture elements i.e. 0.016 cm. let the object-to-hologram distance (s ) be 25 cm, in which case, for a monochromatic o light source of wavelength k = 6328A, the effective hologram width w i l l be about 6.0 cm. Using 525 lines per frame, 30 frames per second, a television camera scanning a hologram of width 6.0 cm has a scanning-spot velocity of approximately v = (525) (30) (6.0) = 94,500 cm/sec. < (2.12) Then from Equation (2.11), the video bandwidth obtained in scanning the hologram is approximately f = S3- = (0.008-) (94.500). 1 \s' (6.328 x 10" 5) (25) - 500 Kc/s . (2.13) 16 The.estimate i s only approximate since Eraunhofer diffraction has been assumed and f^ is seen to depend upon the bar widths, the hologram width and the objeet-to-hologram distance, In the experimental hologram system (see Chapter 4), Eresnel diffraction, 1^ which involves second-*order effects, was used. A theoretical analysis of bandwidth requirements would be much more d i f f i c u l t for this case, but could be expected to show an increase in bandwidth f^ hy a factor of about 2. 2,3.2 Experimental Verification It i s shown in Section 2.4 that we shall be concerned with the holograms of objects consisting of arrays of black and white bars on grey backgrounds,' recorded in the form of photo-graphic transparencies. There are restrictions on the geometry of the optical set-up for Eraunhofer diffraction, as described in Equation ( 2 , 2 ) , to occur, Akhtar has summarized these restric-tionss "The distances of the object from the light source and from the recorded diffraction pattern must be very large compared to the width of the bars in the object." An i n f i n i t e l y large source-to-object distance may be simulated in a f i n i t e distance by using a parallel beam of light to illuminate the object. Calculation shows that for an illuminating beam of parallel red light,' an object-to-diffraction pattern distanoe of about 150 cm w i l l ensure Eraunhofer diffraction for object transparencies in which the bar widths do not exceed 0.4 mm. 17 Pigure 2-6 shows the hologram recorded under conditions of Fraunhofer d i f f r a c t i o n f o r an object consisting of a v e r t i c a l black bar on a transparent background. The hologram i s seen to have the general form predicted by Equation (2.8) from theoreti c a l considerations. (a) (b) Figure 2-6 V e r t i c a l Black Bar (a) and i t s Hologram (b) The significance of this demonstration with regard to t e l e v i s i o n bandwidth-compression i s that the rate of change of intensity from black through grey to white, and white through grey to black i n the hologram,is considerably more gradual than i n the o r i g i n a l object, the v e r t i c a l bar. Thus, the rise-time needed to transmit the video signal obtained by scanning the hologram w i l l be l e s s , which means that the required bandwidth can be reduced. Figure 2-6(b) also shows that the i n t e n s i t y of the higher-frequency components (those farther away from the o p t i c a l axis) decreases rapidly; i n f a c t , beyond a certain point they are not bright enough to r e g i s t e r on a t e l e v i s i o n monitor. ! 8 It is shown experimentally in Section 6 , 1 that a good quality reconstruction can s t i l l he obtained i f only the low-frequency area of the hologram is used. Furthermore, we shall he dealing with oases in which the width of the object transparency is very large compared to the width of any bar of that object, This means that the essential low-frequency area of the hologram due to any bar is small compared to the total area of the holo-gram, and hence, the hologram area to be scanned by the television camera is f i n i t e and of reasonable size, 2,4 Description of the Two-Channel System Akhtar described a possible system which would in theory transmit at a given rate bandwidth-compressed two-tone pictures such as machine drawings and documents. Essentially, the low-frequency components would be transmitted on one channel and the high-frequency components in hologram form on a second channels He predicted 4,5si as a probable bandwidth-compression ratio but no experimental evaluation was attempted," The system could not handle continuous-tone pictures because a continuous-tone picture does not allow transmission- of the spatially-uniform carrier wave required for the Gabor hologram process, A major disadvantage of Akhtar's proposed system is that i t s use is limited to two-tone pictures. The author's system, a modification of Akhtar's proposal, is designed to transmit continuous-tone pictures with reduced bandwidth. As mentioned previously, two channels are necessary because separate coding of the low and high frequencies is to be used. The block diagram (Figure 2-7) shows how this w i l l be done. Transmitter i Receiver Video (a) In B(s) Ed^  Dete< D(. ?e 3 t o r 3) ' (c) Elec. - Opt,. Converter f Optical Eourier Analyser Opt. -Conv - Elec. erter (d) Low-Pass Pilter G(s) Low-Pass Fi l t e r (h) (e) Channel #1 Channel #2 Figure 2-7 Block Diagram of Television Bandwidth-Compression Sys (h) tern (h) Video Out ( g ) Reconstruc-tion F i l t e r H(B) (f) Opt. - Elec. Converter Optical Fourier Synthesizer Elec. - Opt, Converter M(e) B ( s ) « B ( S ) H 20 The waveforms of Figure 2-8 illustrate the principle of the system. These waveforms are obtainable at the points labelled in Figure 2-7. Transmitting End Receiving End Figure 2-8 Waveforms at Labelled Points of the System shown in Figure 2-7 Figure 2-8(a) shows the input signal obtained in scanning a vertical bar. The low-frequency component (b) is transmitted conventionally on channel #1. The edge detector differentiates the video signal with respect to time giving the output (c). Because the video signal was produced by horizontal scanning, \ 21 the edge detector output defines a l l edges not oriented in the horizontal direction. This output can he coded in a number of ways to occupy a smaller bandwidth. In this f e a s i b i l i t y study, the edge-detector output w i l l be displayed on a television monitor giving vertical white and black lines on a grey background. This pattern is photographed and i t s hologram formed optically using a continuous-wave laser. The hologram is scanned with a conventional television camera giving an electrical signal (d) which may be low-pass filtered (e) and transmitted in this form on channel #2. At the receiver, the band-limited hologram is displayed on a monitor, photographed, and i t s reconstruction obtained optically, again using a laser. This reconstruction i s scanned with a television camera giving the waveform (f) which i s seen to again define the edges of the original video signal. The edge signal i s next processed in the reconstruction f i l t e r to give the waveform (g) which when added to the output of channel #1 (b) gives the output (h) as a replica of the input. 22 3 . ELECTRONIC INSTRUMENTATION This chapter deals with the design of the electronic components of the bandwidth-compression system. Because of the need to process video signals without discernible distortion, the design specifications were unusually rigorous. This w i l l be shown in the sections below which cover the design philosophy for each component in turn. 3.1 Low-Pass Filters A low-pass" f i l t e r is required at the transmitting end of channel #1 to obtain the low-frequency components of the input video signal. As w i l l be shown in Section 3«3, an identical f i l t e r is simultaneously required at the receiving . end of channel #2 as part of the reconstruction f i l t e r , Low-pass f i l t e r i n g of the hologram video signal is also necessary before transmission on channel #2. A set of 6 pairs of f i l t e r s , each pair having a different cut-off frequency, was designed and constructed so that the bandwidth of each channel could be varied easily. The minimum bandwidth giving a good quality picture could then be determined readily by subjective testing. Because of the number of f i l t e r s to be built, a simple design was desirable. The type of f i l t e r chosen was governed largely by the need to avoid overshoot because this produces multiple images near edges, a very objectionable form of distortion. This requirement indicated the choice of a f i l t e r hav-ing no overshoot On a step input signal, a requirement f u l f i l l e d by 23 12 the constant-delay or Thomson f i l t e r . A major disadvantage of Thomson f i l t e r s i s that the slope of the attenuation curve just beyond the half-power cut-off frequency i s relatively small unless a large number of inductors and capacitors is used. On the other hand, a maximally-flat-amplitude l ^ (Butterworth )' f i l t e r provides higher attenuation outside the passband, but has considerable overshoot for a step input. •In a 1957 paper by Peless and Murakami,1^ a class of f i l t e r s called Transitional Butterworth-Thomson i s described which offers a compromise, better transient response than the Butterworth f i l t e r and highs' attenuation slope beyond the pass-band than the Thomson f i l t e r . This compromise i s achieved by placing the poles of the Transitional Butterworth-Thomson transfer function Z(s) in the relative s-plane positions shown in Figure 3 - 1 . x Butterworth (maximally-flat amplitude) 0 Thomson (maximally-flat envelope delay) Transitional Butterworth-Thomson Figure 3 - 1 Transfer-Function Pole Positions for Third-Order Fi l t e r s 24 Each Transitional Butterworth-Thomson pole l i e s on a path joining the Butterworth and Thomson poles, and i t s position on which controls the pole angle and i t s distance from the origin, m =; 0 corresponds to the Butterworth pole and m = 1 to the Thomson Pole. Butterworth-Thomson f i l t e r with m = 0.4 would he suitable for the television application. This design would give a calculated overshoot of 3.9$ for a step input and an attenuation of 15.4 dB at twice the cut-off frequency. The f i l t e r s were designed and constructed with cut-off frequencies of 0.25, 0.50, 0.75, 1.0, 1.5 and 2.0 Mc/s. element values for a typical case, a f i l t e r having cut-off frequency of 1.0 Mc/s. The measured frequency response for this f i l t e r i s plotted i n Pigure 3-3. The high-impedance source required for driving the low-pass f i l t e r of. channel #1 is shown in Pigure 3-4. this path may be varied by changing a parameter called 'm' It was decided that a third-order Transitional Pigure 3-2 shows the circuit configuration and -o-18.1 uH W 4.75 — / w -I •o-Pigure 3-2 1.0 Mc/s Low-Pass F i l t e r 25 o 4 8 12 ,o l6 5 20 L <u24 •p -p 28 32 36 40L. * 1 . 3 .4 . 5 1.0 2 Frequency(Mc/s) 4 5 10 Figure 5-5 Frequency Response of 1.0 Mc/s low-Pass F i l t e r -12v Video Input 2.2K-T1 I 100K.ft m 5^ F 2 390-fl Low-Pass F i l t e r G-(s) 2.7KI1-1.5KJI' 50uF 750 pF +6v Note: Transitors type 2N 2495 Figure 5-4 Source for Low-Pass F i l t e r of Channel #1 To Channel #1 Transmitter 26 Video signal from the camera is applied to the high-input-impedance emitter follower T-^ j this is necessary to provide isolation from the differentiator of Pigure 3-7 which i s in parallel with T^. Amplification occurs in the common-emitter stage Tg, and with a 2.2 KT1 load resistor, this transistor appears as a high-impedance source to the low-pass f i l t e r 5(s) which i s terminated in a 7512. resistor. 3.2 Edge Detector The purpose of the edge detector is to locate a l l contours of the input picture not oriented i n the direction of. scanning. Differentiation with respect to time w i l l achieve this. The usually d i f f i c u l t problem of differentiation was made more d i f f i c u l t in this case hy the wide frequency range of •the input signal (30 c/s to 5 Mc/s). An elementary R-C differentiator • was tested and found to yield an unsatisfactorily low output voltage. A signi-ficant amount of noise could he introduced in amplifying such a signal. A simple transistorized differentiator having the circuit diagram of Pigure 3-5 was then constructed. To achieve low input resistance and good high-frequency response, the transistor is operated in common-base mode. is the di f f e r -entiating capacitor and the base-emitter resistance r ^ Q provides the differentiating-resistor. Operating at 2.8 mA emitter current, r, is about 10012. With C, = 47 pP, a low enough 27 -12v R L Output YQ o 2N2495 Input Y±O |j_ °l C 1 = 47 pF R L = 1.8KJ1 2.2ZI1 +6v Figure 3-5 Transistor Differentiator time constant (4.7nsecs) is achieved for differentiation to he nearly ideal at frequencies up to 5 Mc/s. The output signal is taken across the large (1.8 Kfl ) load resistor R^ , thus giving a high-amplitude output. The approximate transfer characteristic (verified hy experiment) is dV Vo = RL°1 ' d T ' ( 3 - D Because of the collector-base capacitance and stray capacitance from collector to ground, high-frequency r o l l - o f f w i l l occur below 5 Mc/s i f R-^  is made too large. This comment applies to a l l circuits in this thesis. The differentiator was tested with square-wave (Figure 3-6(a)) and sinusoidal (Figure 3-6(c)) input signals. The output waveforms, shown in Figures 3-6(b) and (d) respectively, indicate satisfactory circuit operation. F i g u r e 5 - 6 D i f f e r e n t i a t o r Waveforms H o r i z o n t a l s c a l e : 2 0 0 n s e c / d i v i s i o n ; V e r t i c a l s c a l e : 0 . 2 v o l t s / d i v i s i o n ; O s c i l l o s c o p e r i s e t i m e : 3 5 n s e c . (a) Square wave i n p u t ; Cb) Output; (c) S i n u s o i d a l i n p u t (1.2 Mc/s); (d) Output. 29 For reasons described i n Section 3.3 i t was decided that synthesis of the integrating part of the reconstruction f i l t e r would he greatly simplified i f the edge detector were modified from the i d e a l d i f f e r e n t i a t o r form hy shunting the capacitor with a large r e s i s t o r Rg a s shown i n Figure 3-7. • ^ 12v R2 = 47ZH RI = 1 K S 1 5uF m m x l L2 100E-L Video i n C 1 = 47pF 5|iF — / l h - 3ync 0 i n > 2.2KA-4 2.2KJ2 => 560 a T 3 Differentiated Video + S s r n c < > o u t e „ 2 „ 7 K n +6v Notes A l l transitors 2N 2495. Figure 5-7 Modified D i f f e r e n t i a t o r This "bypass r e s i s t o r causes a small percentage of the input signal to appear at the output, hut does not degrade the quality of the holograms formed from the edge-detector output. Synchronizing and blanking pulses from the camera-control-unit must be added to the d i f f e r e n t i a t e d video; t h i s i s achieved by transistors T^ and T^ by virtue of t h e i r common col l e c t o r resistance. The output signal i s taken from the emitter follower T^ which has a suitably low output impedance f o r d r i v i n g a t e l e v i s i o n monitor, through a. 15SI cable. The output video swings 30 about a bias level of 0.6 volts so that when displayed on a monitor, the signal range from black to white i s 0.2 to 1.0 volts, i.e. 0.6 + 0.4 volts, with grey level as 0.6 volts. 3.3 Reconstruction F i l t e r Making the assumption that the hologram construction and reconstruction process has a transfer function of unity, the two-channel system of Figure 2-7 simplifies to that shown in' the block diagram below. Video in B(s) low-'Pass F i l t e r G(B) Edge Detector D(s) Channel # 1 Channel #2 V 1(s) + Reconstruc-tion F i l t e r ' H(s) -s*—o Video out B(B) V 2(s) Figure 5-8 Block Diagram of System Omitting Hologram Process We must synthesize a reconstruction f i l t e r with transfer function H(s) suoh that the overall transfer function for the system is unity. i.e. 1 - G-(s) + D(s) • H(g') (5.2) 31 Therefore i f D(s) i s a differentiaior and G(s) a low-pass f i l t e r , H(s) must perform the operations of integration and high-pass f i l t e r i n g . The block diagram required for synthesis of H(s) is shown in Pigure 3-9. r _ _ _ _ _ _ _ , low-Pass Pi l t e r G(s)-V e - » — o V 2(s) H(s) = V 2(s) High-Pass F i l t e r _ _ i r i - G ( S ) " DTS7 L Figure 5-9 Block Diagram of Reconstruction F i l t e r H(s) It is seen that the low-pass f i l t e r G(s) in the reconstruction f i l t e r of channel #2 is identical to that required at the transmitter of channel #1. The elementary differentiator D(s) shown in Figure 3-10 has transfer function V R,C,s D(s) = ^  = R 1C 1s + 1 (3.4) 52 R-, V Figure 5-10 Elementary Differentiator. Therefore, a complementary integrator would have transfer function DlsT R-^s + 1 " R ^ s (3.5) This function has a pole of transmission at zero frequency and is therefore not practically realisable. To overcome this d i f -ficulty the differentiator has been modified to have the following equivalent circuit. R„ + + V , R. Figure 5-11 Modified Differentiator The transfer function for this circuit is 33 R ^ s + R-^ Rg D ( s ) = R 1C 1s + (R-^ /Rg + 1) (3.6) •Comparing Equation (3.6) with Equation (3.4) we see that i f Rg >^ Rj_ the networks of Figures 3-10 and 3 - H w i l l behave the same at high frequencies,' while at low frequencies, a small fraction (—R-^ /Rg) of the input signal w i l l appear at the output of the modified circuit. Prom the transfer function of the modified differen-tiator (Equation (3.6)), a complementary "integrator" would have transfer function 1_ R. .JC-JS + (R^Rg + 1) R 1C 1s + R]_/Rg (3.7) At zero' frequency, transmission is now fi n i t e , and the transfer function l/D(s) i s realizable in the following form. - o - > -0 I a V o o f • Figure 3-11 R, C, d = R. Figure 5-12 Integrator Complementary to Differentiator of Figure 5-11 If the current?' I is made proportional to the output voltage from the differentiator, the element values R^ , Rg, and C^ have the same values as used in the differentiator. The 54 practical circuit was realized by using a wide-band transistor in common-base mode as a current source and the combination of R^ , Rg, and C^ as the collector load (Pigure 3-13) . R 1 = 100ft C 1 = 47pF -]" Input •12Y Integrated Output +6v Rote: Transistors type 2N 2495. Pigure 5-15 Integrator Circuit The integrating transistor T^ i s followed by an emitter follower Tg so that the collector circuit Of the integrator i s not loaded. When tested, the differentiator-integrator combination was found to have a transfer function of 0.07 + 2$ at a l l frequencies up to 5 Mc/s. The remainder of the reconstruction f i l t e r is instru-mented as shown in Pigure 5-14. The input signal from the inte-grator is amplified in the common-emitter stage T^. One output from the emitter follower Tg is further amplified and phase inverted in T^ which is also the current source required for driving the low-pass f i l t e r G-(s). The filtered signal, after isolation provided by the emitter follower T^, is applied to the emitter of T ^ The other output from Tg is applied to Tg^ As - 12 v Input from Inte-grator 150SL pw-|H • 330SL '« < ~ ^ ^270 ft 4O11F ^ 2.7kJ\ « 50uF Note: A l l Transistors Type 2N2495 Figure 5-14 Part of Reconstruction F i l t e r V>1 From Channel # 1 36 indicated in Figures 3-4 and 3-14? the low-frequency signal from channel #1 reaches the adder through T^. Signal addition is performed in T^ ., Tg and T^ because the transistors have a common collector resistance. The f i n a l output signal i s taken from the emitter follower T g which provides a low output impedance for driving a monitor through, a 75 SI cable. 3.4 Tests of the Electronic Instrumentation The components described in the previous sections of this chapter were assembled in the positions shown in the system block diagram (Figure 2-7). To test the system operation, a pulse input signal was applied and the waveforms monitored at various points. The photographs of these waveforms, shown in Figure 3-15, correspond to the expected waveforms as sketched in Figure 2-8. The low-pass f i l t e r used for the photographs had a cut-off frequency of 0.5 Mc/s. The output signal (Figure 3-15(f)) shows evidence of a slight overshoot but otherwise closely resembles the input. 57 Figure 5-15 System Waveforms for a Pulse Input Signal Horizontal scale: 200nsec/division; Vert ical scale: 0.2volt/division; Oscilloscope rise-time: 22nsec. fa) Pulse inputj (b) Output from 0.5 Mc/s low-pass f i l t e r ; (c) Output from edge detector; (d) Input to reconstruction f i l t e r ; fe) Output from reconstruction f i l t e r ; (f) Output (obtained "by superimposing waveforms (b) and (e)). 38 4. OPTICAL INSTRUMENTATION Akhtar has described in detail the procedure for making holograms of two-tone line drawings, and for forming 15 the reconstructions from these holograms» The author has adapted the techniques to make them directly applicable to the bandwidth-compression system being studied in this thesis, A description of these and other techniques used in the system is given in this chapter, 4°1 Hologram Formation from Edge-Detector Display The edge-detector output when displayed on a tele-vision monitor has the form of black and white lines on a grey background, A number of processes are necessary to form the hologram of this edge pattern. In the, f e a s i b i l i t y study, the monitor display was photographed with a Voigtlander Vito II 35mm camera. This camera has an iris-type shutter? a sliding-door (focal plane) -type shutter in conjunction with the sequential television scanning pattern was found to produce unequal exposure across the width of the film. Best results were obtained with Kodak M417 high-contrast copy film? a typical exposure was 0.5 sec. at an aperture of f3»5» The contrast of the monitor display was found to be considerably enhanced i f the room was completely darkened. An example of an edge-pattern photograph is given in Figure 6-1(c). 39 After development, the hologram of the edge pattern was formed from the film transparency using the optical arrangement shown in Pigure 4-1» 10 cuH CW Laser Lens Pinhole Object t(x,y) Hologram Vidicon Target Plate 1 Pigure 4-1 Experimental Arrangement for Recording Holograms Lasers Spectra-Physics Model 112, 10 mw output power, 6328A wave length (red), 4°5mm heam diameter. Lens: f2.2, 50mm focal length. Pinhole; 220 microns diameter. It is seen that the hologram is formed directly onto the target plate of the television camera? this produces a high signal-to-noise ratio» The pinhole acts as a low-pass 15 f i l t e r to remove the Airy rings produced by the lens. A l l equipment was rig i d l y mounted on an optical hench as shown in the photograph of Pigure 4-2. 40 Television Figure 4-2 Photograph of Hologram Recording Apparatus It may he noted in Figure 4-1 that the illuminating light heam is divergent rather than parallel, and that the hologram is formed at a relatively short distance from the object transparency. Such an arrangement was necessary so that at a later stage the hologram reconstruction could be formed at a workable distance from the hologram. The result i s Fresnel rather than Fraunhofer diffraction; the general arguments used in Chapter 2 are s t i l l valid, although the mathematical expressions are much more complicated for Fresnel than for 1 fi Fraunhofer diffraction. The video signal obtained in scanning the hologram was low-pass filtered as discussed in Section 2.3.2 and transmitted in this form on channel #2. 41 4o2 Hologram Reconstruction Process At the receiver of channel #2, the low-pass filtered video signal from the hologram was displayed on a television monitor and photographed on Kodak TRI-X 35mm film; an example of such a photo is given in Figure 6-l(e). The hologram reconstruction was formed from the transparency using the arrangement of Pigure 4-3 • 100 ci 1-0 cm-H Reconstruction t(x,y) Lens Pinhole Hologram U (to , co ) x y y Figure 4 - 3 Experimental Arrangement for Obtaining Reconstruction Because of i t s large area the reconstruction could not be formed directly onto a Vidicon target plate. A photograph was therefore taken of the reconstruction and printed on high-contrast paper. Pigure 6-l(f) gives an example of such a reconstruction. The hologram could have been focussed onto a Vidicon tube i f a wide-aperture lens had been available. Alternatively, an Image-Orthicon camera tube, which has a diameter of 4 inches, could have been used. 42 The edge-pattern reconstruction was then scanned with a t e l e v i s i o n camera and the r e s u l t i n g video signal was applied to the input of the reconstruction f i l t e r ? a photograph of the output i s shown i n Figure 6-1(g). 4*3 Real-Time Recording of Holograms•and Reconstructions It can he seen from the descriptions given i n Sections 4°1 and 4»2 that the procedure used to form the hologram from the o r i g i n a l video signal and subsequently to convert the hologram reconstruction into a video signal i s an involved process and at present cannot be performed i n r e a l -time » For the purpose of a f e a s i b i l i t y study t h i s i s not a great disadvantage except that i t precludes the p o s s i b i l i t y of tes t i n g the o v e r a l l system with anything except s t i l l pictures. To perform the hologram operations i n real-time we must image the edge pattern, as displayed on a t e l e v i s i o n monitor, onto a "screen" whose o p t i c a l transmission at any point i s proportional to the l i g h t i n t e n s i t y at that point, i . e . , l i k e a photographic transparency* Then when the screen i s temporarily illuminated by a laser beam, the required hologram would be formed on the target plate of a t e l e v i s i o n camera situated behind the screen. A similar process i s needed f o r the reconstruction. 43 5 . ALTERNATIVE, BANDWIDTH-COMPRESSION SYSTEMS USING EDGE-CODING It was anticipated that the two-channel compression system using hologram transmission would have some serious practical disadvantages, and some thought was consequently given to possible alternative approaches to the'.''coding problem. As a result , two alternative systems were devised and are described in this chapter. These systems resemble the hologram system in that they code the edge pattern obtained by differentiating the input video signal with respect to time, but have the advantage of avoiding the electro-optical transformations essential to.the hologram method. 5 . 1 Practical Disadvantages of the Hologram Transmission System Although i t was shown in Section 2.3 that the proposed bandwidth-compression system using hologram transmission is attractive from bandwidth considerations, the following disadvan-tages in a practical instrumentation were anticipated: (a) At present the system cannot be instrumented for real-time operation. At the transmitting' end this • is due to the impossibility of forming the hologram of the edge pattern directly from the televised scene. Similarly, at the receiver, the hologram reconstruction cannot be formed directly from the video signal of the hologram,; This aspect was discussed more fu l ly in Section 4.3. 44 (b) The quality of the Gabor hologram reconstruction is impaired hy the presence of the conjugate image (see Section 2,2). Reduction of the impairment involves relatively complicated optical processing such as recording a second hologram in a plane separated from the f i r s t hologram plane by an odd number of quarter-wavelengths, bleaching this hologram, and then superimposing on the f i r s t holo-gram before taking the reconstruction. (c) Due to spatial non-linearity in the electro-optical conversion processes, registration of the images from the two channels over the whole raster area would be almost impossible. Under the best attainable working conditions, television cameras and monitors are at least 1$ spatially non-linear. Consequently, because channel #1 of the proposed hologram transmission system uses one electro-optical conversion, and channel #2 uses five electro-optical.conversions, registration of the two images can only be expected over a small fraction of the raster area. This is a serious disadvantage because registration errors are very easily noticed. Bearing these disadvantages in mind, two alternative bandwidth-compression systems have been devised. These systems can be instrumented for real-time operation and should give better quality pictures than the hologram system. 4 5 5 . 2 An Alternative Two-Channel .System using Edge-Coding V. A two-channel system which avoids holograms and conse-quent electro-optical conversions at intermediate stages and which is capable of real-time instrumentation is shown in Pigure 5 - 1 . This system achieves bandwidth compression by taking advantage of the high degree of vertical correlation which exists in the edge pattern obtained by differentiating the input scene bright-ness pattern with respect to horizontal position. Transmitter Receiver Video in (a) B(t) Low-Pass' F i l t e r G K s ) Edge Detector D(s) Edge Coder. (b) ^ (f) Video out < + > T ^ B (t) - B(t) (e) Reconstruct tion F i l t e r HifJ Edge Decoder A ( d ) Figure 5 - 1 Block Diagram of System using Edge-Coding 46 Assuming that the original video signal B(t) is produced by conventional horizontal scanning (see Pigure 2-1), the edge pattern, which is obtained hy differentiating B(t) with respect to time, is stored temporarily as a two-dimensional 17 brightness pattern in a video storage device. An output signal is obtained using a vertical scan for read-out from the storage device, with simultaneous erasure of the stored information. This signal is then low-pass filtered and trans-mitted on channel #2, By similarly writing and reading in orthogonal directions at the receiver decoder, a close resemblance to the original edge pattern is obtained. This provides the input signal to the reconstruction f i l t e r , which, as shown in Section performs the operations of integration and high-pass f i l t e r -ing. The output from the reconstruction f i l t e r is added to the low-pass filtered version of B(t) (transmitted on channel #l), giving a bandwidth-compressed picture with edge detail clearly defined. Edge definition is controlled by the cut-off frequencies of the low-pass f i l t e r s . This system is seen to require temporary video storage at the coder and decoder of channel #2 during processing. Because a suitable storage device is not available at this laboratory, a real-time simulation of the system could not be carried out, but tests to verify the practicability of the scheme are reported in Chapter 6. 47 5«3 Basis for a Proposed One-Channel System using Edge-Coding Figure 5-2(a) shows a typical television scene and Pigure 5-2(h) i s the edge-pattern obtained by differentiating the spatial brightness of the input scene with respect to horizon-ta l position. It is apparent that edges not oriented in the direction of scanning occupy only a small fraction of the total raster area and exhibit a high degree of spatial correlation. Consequently, i f data giving the position and brightness of these edges can be coded to give a nearly constant transmission rate, the required channel capacity should be considerably re-duced,1 Assuming the edge pattern to be produced at the transmitter by differentiating the video signal with respect to time, then a complementary integrator at the receiver would give a close likeness to the original signal. Suitable circuits for the differentiating and integrating devices are described in Chapter 3» This system can be instrumented for real-time operation and could conveniently use digital data transmission, an important consideration in view of present-day trends. The system can be extended to incorporate frame-to-frame correlation, thus increasing the compression ratio. It should also be possible to take advantage of the small rate of data absorption of the human eye-brain recognition process to reduce .the high data-transmission rate, which normally occurs during periods of scene changes, camera panning and object movement. Such a system appears attractive for the television-phone application as well as in television relay links. 48 Figure 5-2 Test Scene (a) and i t s Edge Pattern (b) 4 9 6. TEST RESULTS The basic bandwidth-compression system using hologram transmission of the edge information and the alternative system: described .in Section 5.2 were tested using s t i l l slides of typical television material. The results obtained are described in this chapter*, 6*1 The Two-Channel--System using the Hologram'Principle The equipment for the system was assembled as shown in the block diagram of Pigure 2-7* A test scene was scanned with a studio-quality Marconi Mark-4 Image-Orthicon television camera system having a bandwidth of 6Mc/s, The output video signal ' was differentiated using the circuit of Pigure 3-7 and the ' resultant edge pattern displayed on a 14-inch Conrac monitor which had previously been adjusted for good linearity. The edge pattern was photographed with a 35mm camera and the holo-gram of this edge pattern was formed on the target plate of a ' Vidicon television camera tube as described in Chapter 4» The video signal obtained in scanning the hologram was low-pass fi l t e r e d , transmitted on channel #2, and the display on a monitor at the receiving end was photographed. The latter procedure was repeated using the set of third-order Transitional Butterworth-Thomson f i l t e r s described in Section 3»1* The • cut-off frequencies of the f i l t e r s were 0.25, 0,50, Ow75, 1.0, 1.5, and 2.0Mc/.s»., The bandwidth-limited holograms were then reconstructed using the apparatus shown in Pigure 4~3» 5 0 As expected from the theory of the hologram reconstruc-tion process developed in Section 2.2, the quality of the reconstructions was impaired hy the presence of the conjugate image which forms behind the real image. The conjugate image, being out of focus, provides a noisy background to the real image. In the G-abor-type hologram used here these impairment 5 effects are not easily reduced. The hologram reconstructions were inspected and i t was found that holograms transmitted with bandwidths as low as 0.75Mc/s gave edge pattern reconstructions of reasonable quality, although impaired by the presence of the conjugate image. The edge pattern reconstructed from the 0.75Mc/s bandwidth hologram was then "scanned with a television camera and the resultant 1 video signal applied to the input of the reconstruction f i l t e r * Simultaneously,, the video signal obtained in scanning the test scene with the Marconi camera was low-pass fil t e r e d , transmitted on channel #1, and added to the output of the reconstruction f i l t e r giving a picture resembling the test scene* Low-pass f i l t e r s having cut-off frequencies ranging from 0.25 to 2.0Mc/s were used in turn to limit the bandwidth of the picture transmitted on channel #1. The series of photographs in Figure 6-1 shows the processing of a test scene at different stages of the system. For the case shown,- the hologram was transmitted with a bandwidth of 0.75Mc/s and the low frequency information with a bandwidth of 0.5Mc/s. 51 (a) (e) (b) (f) (c) (a) (h) Figure 6-1 Feasibility Study Results for Hologram System ]a) Normal picture (6Mc/s bandwidth) i ,b) 0.5Mc/s bandwidth picture; (c) Edge pattern; ,d) Hologram of edge pattern (6Mc/s bandwidth)) (e) 0.75Mc/s hologram; (f) Reconstruction of edge pattern from 0.75Mc/s holo-gram ; (g) Output from reconstruction f i l t e r ; (h) Bandwidth-compressed picture (total transmission bandwidth 1.25Mc/s). 52 Two defects of the system can he seen by comparing the original and f i n a l pictures (Figures 6-1(a) and (h) respectively)? (i) The conjugate image of the hologram reconstruction (Figure 6-l(f)) produces uneven background noise in the f i n a l picture? this has the effect of blurring the fine detail (as for instance, where there are windows in the buildings). It was found, that even with no compression, this defect i s apparente ( i i ) Registration of the superimposed pictures could not be attained over the whole rast % for the case shown this has caused horizontal distortion and made the edges in the l e f t half of the picture appear blurred. . This defect was anticipated (Section 5.1) because a l l stages of the system are ! spatially non-linear. The principal sources of non-linearity are the electro-optical conversions produced by the television cameras and monitors, and slight curvature of the films used for recording the holograms and their reconstructions* Spatial non-linearity in the electro-optical conver-sions i s particularly serious because the system used for the f e a s i b i l i t y study involved six of these conversions, and furthermore, the f i n a l picture was obtained by superimposing the video signals from two television cameras having different non-linearity properties* 53 Prom these tests i t may he concluded that although the concept of a television system which transmits the picture edge information in hologram form i s theoretically attractive, practical disadvantages are caused hy the conjugate image of the hologram reconstruction and the problem of superposition? no simple solution exists to either problem. In addition, a real-time instrumentation of the system is not possible at present. 6.2 The Alternative Two-Channel System using Edge-Coding The equipment for the system was assembled as' shown in the block diagram of Figure 5-1* Because a suitable video storage device was' not available at this laboratory, a real-time simulation of the system was not possible. Consequently, for the purpose of a f e a s i b i l i t y study, the edge pattern obtained by differentiating the input video signal was displayed on a monitor and recorded photographically. This pattern was then rotated through 90° and scanned with a conventional television camera*- After low-pass f i l t e r i n g , the resultant video signal was transmitted on channel #2 and a similar procedure adopted at the receiver to retrieve the edge pattern. The video signal obtained in scanning this edge pattern was applied to the reconstruction f i l t e r and the output added to the low-pass filtered version of the original video signal which was trans-mitted on channel #1, Low-pass f i l t e r s having cut-off frequencies ranging from 0*25 to 2„0Mc/s were used in turn to limit the bandwidth 54 Figure 6-2 Test Results for Two-Channel Edge-Coding System (a) Normal picture (6Mc/s bandwidth); (b) 0 . 5 M c/s bandwidth picture; (c) Edge pattern (6Mc/s bandwidth); (d) Edge pattern scanned in vertical direction ( 0 . 5 M c/s bandwidth); (e) Output from reconstruction f i l t e r ; (f) Bandwidth-compressed picture (total bandwidth l.OMc/s); 55 of the two channels. The series of pictures shown in Figure 6-2 illustrates the operation of the system for the case where both cannels had a bandwidth of 0.5Mc/s giving a compression ratio of about 6:1. Registration errors in the superimposed picture (Figure 6-2(f)) are caused by spatial non-linearity in the television cameras and monitors. Registration would not. be a serious problem in a real-time system using a video storage device at the edge-coder and decoder of channel #2 as described in Section 5.2. On the basis of the results shown in Figure 6-2, a real-time, a l l - e l e c t r i c a l system could be expected to give a higher bandwidth-compression ratio than 6:1. Such a system would allow better u t i l i z a t i o n of existing microwave and coaxial cable television relay links. Assuming a compression ratio of 6:1, the bandwidth required for transmission of commercial quality television would be about 800Ec/s. 56 7. CONCLUSIONS It has been shown that the bandwidth required for television transmission can be reduced by using a system which transmits the low-frequency information by conventional means on one channel, and the high-frequency information in hologram form on a second channel., Experimental tests indicated that the system had the following disadvantages: (i) It could not be instrumented for real-time operation,. ( i i ) The out-of-focus conjugate image of the hologram reconstruction impairs the quality of the fi n a l picture by blurring the fine detail, ( i i i ) Registration of the images from the two channels could only be achieved over a small fraction of the raster area. This was caused by spatial non-linearity in the television cameras and monitors used for electro-optical conversions at the intermediate stages of the system* Impairment produced by the defects ( i i ) and ( i i i ) mentioned above prevented satisfactory quality judgementsfrom being made, but a bandwidth-compression ratio of about 4.8 s i , in approxi-mate agreement with the ratio theoretically predicted by Akhtar, was indicated. Consequently^ an alternative two-channel bandwidth-compression system using edge-coding of the high frequencies was 57 proposed. This system could he instrumented for real-time operation and avoided the other disadvantages of the hologram system. Due to the lack of video storage equipment, electro-optical conversions and photographic techniques, both of which degrade the f i n a l picture quality, had to he used at intermediate processing stages in the f e a s i b i l i t y study. The system was found to provide an efficient method of separately transmitting the low- and high-frequency information and gave a bandwidth-compression ratio of about 6:1. A higher ratio could be obtained in a real-time a l l - e l e c t r i c a l system. A one-channel real-time system which transmits only the edge information, with consequent reduction in bandwidth, was also proposed. An advantage of the system is that i t could be extended to take advantage of frame-to-frame correlation and the psychophysics of vision to further reduce the transmission channel bandwidth requirements. On the basis of these results i t is f e l t that the hologram system is not practically feasible. However, i t is believed that the other two systems proposed in this thesis could be developed to give relatively simple methods of obtaining a reduction of about 6:1 in bandwidth requirements for either black and white or colour television transmission. 58 REFERENCES Teer, K., 11 Investigations into Redundancy and Possible Bandwidth Compression in Television Transmission," Philips Research Report, 1 4 , No. 6 , Dec, 1959? 1 5 , No.l, Feb., I960. Shannon, C.E., "A Mathematical Theory of Communication," B.S.T.J.. 27, 1 9 4 8 , p. 3 7 9 . Seyler, A.J., "The Coding of Visual Signals to Reduce Channel Capacity Requirements," Proc. IEE, 109, C, Sept., 1 9 6 2 , p. 6 7 6 . Seyler, A.J. and Budrikis, Z.L., "Detail Perception after Scene Changes in Television Image Presentations," IEEE Trans, on Information Theory. IT - 1 1 , Jan., 1 9 6 5 , p. 3 1 » Pan, J.W., "Picture Processing," Quarterly Progress Report, No. 6 6 , M.I.T. Research laboratory of Electronics, July, 1 9 6 2 , p. 2 2 9 . Huang,. T.S., "PCM Picture Transmission," IEEE Spectrum. 2 , Dec, 1 9 6 5 , p. 62. Gabor, D., "Microscopy by Reconstructed Wavefronts," Proc Roy. Soc. A, 197, 1949, p. 4 5 4 . i Leith, E. and Upatnieks, J., "Wavefront Reconstruction with Diffused Illumination and Three Dimensional Objects," J. Opt. Soc Am.. 5 3 , Nov., 1 9 6 4 , p. 1 2 9 5 . Leith, E„, "Requirements for a Wavefront Reconstruction Television Facsimile System," J.S.M.P.T.E.. Oct., 1 9 6 5 , p. 893» Born, M, and Wolf, E., Principles of Optics. 2 n d . edition, New York; Pergamon Press, Inc., 1 9 6 5 , p* 3 8 5 . Akhtar, S.A., "Video Bandwidth Compression Using Hologram Technique," M.A.Sc Thesis. Department of Electrical Engineering, Faculty of Applied Science, The University of British Columbia, August, 1 9 6 5 , p. 26. Ibid., p. 1 2 . Ibid., p. 28. Thomson, W.E., "Networks with Maximally Flat Delay," -Wireless Engineer. 29» 1 9 5 2 , p.. 2 5 6 . 59 13 • Butterworth, S,, "On the Theory of Amplifiers," Exp. Wireless. 2> 1930, p. 536. 14. Peless, Y. and Murakami, T., "Analysis and Synthesis of Transitional Butterworth-Thomson Filters and Bandpass Amplifiers," R.C.A. Review. 18, 1957, p. 60. 15. Akhtar, S.A., op. c i t . , p. 32. 16. Born, M. and Wolf, E., op. cit», p. 383. 17» Potter, J.B., "On the Use of the Vidicon Camera Tube as a Video Storage Device," Proc. IRE (Aust.). 24, Dec, 1963, P» 855. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0103228/manifest

Comment

Related Items