UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Image processing Chan, Andy Bing-Bill 1995

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
[if-you-see-this-DO-NOT-CLICK]
ubc_1995-0187.pdf [ 4.78MB ]
[if-you-see-this-DO-NOT-CLICK]
Metadata
JSON: 1.0086896.json
JSON-LD: 1.0086896+ld.json
RDF/XML (Pretty): 1.0086896.xml
RDF/JSON: 1.0086896+rdf.json
Turtle: 1.0086896+rdf-turtle.txt
N-Triples: 1.0086896+rdf-ntriples.txt
Original Record: 1.0086896 +original-record.json
Full Text
1.0086896.txt
Citation
1.0086896.ris

Full Text

IMAGE  PROCESSING by  A N D Y BING-BILL C H A N B . S c , The University of B r i t i s h C o l u m b i a ,  1993  A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T THE REQUIREMENTS  FOR THE DEGREE  M A S T E R OF  OF  SCIENCE  in T H E F A C U L T Y OF G R A D U A T E STUDIES Department of Statistics  We accept this thesis as conforming to ^he required standard  T H E UNIVERSITY OF BRITISH COLUMBIA April  1995  © A n d y Bing-Bill Chan,  1995  OF  In  presenting  degree freely  at  this  the  available  copying  of  department publication  of  in  partial  fulfilment  of  the  University  of  British  Columbia,  I  agree  for  this or  thesis  reference  thesis by  this  for  his thesis  and  scholarly  or for  her  of  T h e U n i v e r s i t y o f British Vancouver, Canada  Date  DE-6  (2788)  Columbia  I  further  purposes  gain  that  agree  may  be  It  is  representatives.  financial  permission.  Department  study.  requirements  shall  not  that  the  Library  an  granted  by  allowed  advanced  shall  permission  understood be  for  the that  without  for head  make  it  extensive of  my  copying  or  my  written  ABSTRACT  In this thesis, we consider the restoration of multiple grey levels image. The problem is to clean up or restore the dirty picture, that is, to construct an estimate of the true image from a noisy picture of that true image. Following a method proposed by Meloche and Zamar (1994), we estimate the colour at each site by a function of the data available in a neighbourhood of that site. In this approach, the local characteristics of that image, that is, the frequency with which each pattern appears in the true unobserved image are particularly important. We will propose a family of unbiased estimates of the pattern distribution and the noise level which are used in the restoration process. We will use our estimates of the pattern distribution in an attempt to select the best neighbourhood shape for the restoration process.  ii  CONTENTS Abstract  "  Table of Contents  iii  List of Tables  iv  List of Figures  v  Acknowledgements  vii  1 Introduction  1  2 Estimations of Q  10  2.1 Introduction  10  2.2 Unbiased Estimations of Q  10  2.3 A Simple Representation for Q  21  3 Joint Estimations of Q and a  29  3.1 Introduction  29  3.2 Estimation of a using Estimating Equation derived from E{\ £" *i ) 2  2 9  =1  3.3 Estimation of a using Estimating Equation derived from E(\  J2i=i  3  4 Neighbourhood Shapes  3  5 Conclusions  3  3  5  References  §  40  in  LIST OF TABLES Table 1.1 Frequency distribution (r = 1)  5  Table 1.2 Frequency distribution (r = 2)  5  Table 1.3 Frequency distribution (r = 5)  7  Table 2.1 Pattern distribution with r = 1, <p(8,y) Table 2.2 Pattern distribution with r  =  =  Indicator functions  1, <p(8,y) = Power functions  25 25  Table 2.3 Pattern distribution with r = 2, ip(6, y) — Indicator functions  26  Table 2.4 Pattern distribution with r = 2, o?(<5, y) = Power functions  27  Table 4.1  AMSE  for various neighbourhood sizes and shapes  iv  .36  LIST OF FIGURES  Figure 2.1 True Image (Unobserved)  41  Figure 2.2 Noisy Image (Observed)  41  Figure 2.3 Pattern distribution with r = 3  42  (i) <p(8, y) = Indicator functions  42  (a) Q (8) vs«5  42  (b) Q (8)  42  3  3  vs8  (c) Q {8)vs8  :..  3  (ii) <p(8, y) = Power functions (d) Q (6)  43  vs8  43  Q (8) vs8  43  3  (e)  3  (f) Q(<5)vsc5  43  3  Figure 2.4 Pattern distribution with r = 5 (i) tp(6, y) = Indicator functions (a)  42  44 44 44  Q (8) vs8 3  (b) Q (8)  vs8  44  (c) Q (8)  vsS  44  3  3  (ii) (p(8, y) = Power functions (d) Q {8) 3  45 45  vs8  (e) Q (8) vsc5  45  3  v  (f) Q (6)vs6  45  3  Figure 2.5 Restored Image  46  (i) with (f(8, y) = Indicator functions (a) r  46  1  46  (b) r = 2  46  =  (c) r = 3  .'  46  (d) r = 5  46  (ii) with (f(6, y) — Power functions (e) r = 1  47  ^  47  (f) r = 2  47  (g) r = 3  47  (h) r = 5  47  Figure 3.1 True Image (0-1-2 strips)  48  Figure 3.2 Noisy Image (0-1-2 strips)  48  vi  ACKNOWLEDGEMENTS  I would like to thank my supervisor, Jean Meloche, for all his support, encouragement, valuable time, and excellent supervision he has provided for the development of my thesis and my thesis presentation on April 4th 1995. I am honoured to be the first graduate student to work on research and thesis under his supervision. I was introduced to the field of Image Processing by Jean Meloche in the fall of 1993. I am amazed by the fact that statistics can be widely applied to Image Processing. It has always been fun working with Jean, and I have learned a lot from him. He has been one of the greatest teachers whom I have ever met. Without him, this thesis may never be made possible. I would also like to thank Ruben Zamar for all his wonderful and helpful comments, and the time he spent with me on the directed study course. Special thanks to my two special friends, Grace Chiu and Andy Ho, for their help in my C programming, and all the fellow graduate students for making my life in the Department of Statistics exciting.  vii  1 Introduction.  In this thesis, we consider multiple grey levels image restoration. One kind of grey levels is grey levels from white to black. Some other kinds are grey levels are from white to red, white to green, white to blue, etc. If we combine the three latter kinds, we will get a coloured image. The problem is to clean up or restore the dirty picture, that is to construct an estimate of the true image from a noisy picture of that true image. The true image is not observed, while the noisy picture is. Figure 2.1 is the true image while Figure 2.2 is a degraded version of true image, the noisy picture. We want to recover the true image (Figure 2.1) from the noisy picture (Figure 2.2) which is the observed data. The estimate of the colour at each site proposed by Meloche and Zamar (1994) is a function of the data available in a neighbourhood of that site. The local characteristics of the underlying image is central to their approach. The notion of local characteristics is formalized in what follows. We assume that there are n sites or pixels on a plane (or a line), numbered 1 to n. The vector {zi, z^,..., z } represents the true image, and 2,- € C where C = {ci, C 2 , . . . , cjt} n  is a set of k grey levels. The observations Yi, Y ,..., Y (called the records) form the 2  n  degraded image and are modelled as (1)  Y = z + ae i  i  i  where e\,..., e„ i.i.d. N(0,1), Yi,..., Y are the records, and cre.'s are the noise. The obn  jective of image restoration is to work out estimates z\,..., z of the true image z\,..., z n  n  from the records Y\,..., Y . n  Following Meloche and Zamar (1994), we define a system of neighbourhoods, Ni,..., N . n  1  The neighbourhood Ni, centred at i, consists of r sites called the neighbours of site i and includes the site i itself. In general, the neighbourhood JV,- is the set of all sites j at a small distance from site i. The neighbourhoods N\,...,  /V have the same shape but n  different centres. For examples, if /Y, = {l,...,n} for all i, then r equals n and we have just one neighbourhood, the whole image. On the other hand, if N{ — {i} for all i, then r equals 1 and then we have n single-site neighbourhoods. Neighbourhoods can be 1-dimensional (linear), 2-dimensional (planar), 3-dimensional, or even higher than 3-dimensional. A pattern is a particular colouring of the sites (or pixels) in a neighbourhood. Some examples of neighbourhoods are graphically shown in the following examples. Example 1.1 : If the true image consists of 0 — 1 bits sent over a line, then the following is a typical linear 3-site neighbourhood:  The true image is one-dimensional, and the colour set C is {0,1}, where 0 = black, and 1 = white. In this example of 3-site neighbourhood, there will be 2 = 8 possible 3  patterns:  Mathematically, a pattern here is a vector of 3 coordinates of black or white that represent the colour of neighbourhood sites in somefixedorder.* As we will notice from the next example, the size of the pattern space increases rapidly with the size of the neighbourhood. Example 1.2 : Suppose we have a colour set C = {0,1,2} (0 = black, 1 = middle grey, 2  2 = white), and a cross-shaped neighbourhood  There will then be 3= s  0 0 0 0 0  0 0 0 0 1  0 0 0 0 2  Again, mathematically, a pattern here is a vector of 5 coordinates of black, middle grey or white that represents the colour of neighbourhood sites in some fixed order.* There is a need for a systematic convention for the ordering of the components of a neighbourhood. Firstly, we denote a pattern by 8. We use the bold face 6 as a vector of <5's in a neighbourhood. In Example 1.1, 6 is a vector of 3 coordinates, that is, 6 — (6"i, <5, 83), 2  and in Example 1.2,  6  is a vector of 5 coordinates, that is, 6 =  (Si, 82, S3,84,6*5).  In the  case of 3-site neighbourhoods, we use 6 = (Si, <5, 83) in the following order: 2  <5i And then in the case of cross-shaped neighbourhoods, we use 6 =  (81,62,83,84,85)  in the  following order:  82  83  84  Ss  Bold face notation is used to represent a vector of variables in a neighbourhood. The coordinates are ordered in the same way as the coordinates of the patterns are. For  example, for 3-site neighbourhoods, Y,- =  (^,1,5^,2,^,3)  is the vector of records, Yj, for  j in the neighbourhood Ni of site i. Yij is the record at the j  th  neighbour of site i. By  equation (1), Yi.i  =  Yi,2  i,l  z  i,2  z  Zi,3  +  cre  it2  o-e  i<3  so that the distribution of Y - is multivariate normal with mean z,- and covariance matrix t  <7 / . 2  3  Throughout this thesis, we look at the distribution of all the possible patterns in the true image. The true distribution of the patterns is usually unknown, but it can be estimated from the records (the noisy picture) in many different ways. As stated before, the pattern distribution plays a useful role in the process of image restoration. The estimates z\,...,z of the true image proposed by Meloche and Zamar (1994) are n  based on the knowledge of the pattern distribution and cr. Example 1.3 : In this example, we use the Figure 2.1 with 3 grey levels as the true image. Let r denote the neighbourhood size, and C = {0,1,2}. When r = 1, there are only three possible patterns, namely, black, middle grey, and white. Table 1.1 below shows how frequently each pattern appears in the true image.  4  Table 1.1: Frequency distribution (r = 1) pattern frequency black  0.48340  grey  0.42505  white  0.09155  Total  1.00000  This frequency distribution reflects the fact that 48% of all pixels are black, 43% of them are middle grey, and 9% are white. When r = 2, a horizontal 2-site neighbourhood, there is a total of 3 = 9 possible patterns, namely, 00, 01, 02, 10, 11, 12, 20, 21, and 22. 2  Table 1.2 below shows the frequency with which each pattern appears in the true image. Table 1.2: Frequency distribution (r = 2) pattern frequency 00  0.47192  01  0.00842  02  0.00305  10  0.00842  1 1  0.41541  12  0.00122  20  0.00305  21  0.00122  22  0.08728  Total  1.00000  5  The high frequency of the patterns 00, 11, and 22 in this frequency distribution reflects the fact that neighbouring pixels typically have the same colour. When r = 5, we consider the cross-shaped neighbourhood, there are 3 = 243 possible 5  patterns. We use 6 = (6"i,  82,83,84,  8) 5  in the following order:  Table 1.3 below is the corresponding frequency table.  6  Table 1.3: Frequency distribution (r = 3) pattern frequency  pattern frequency pattern  0 00 0 0 0.43811  10 0 0 0 0.00824  2 0 0 00  0.00293  0 00 0 1 0.00824  10 0 1 0 0.00006  20 0 1 0  0.00012  0 00 0 2 0.00293  10 0 20  0.00012  20 2 2 0  0.00018  0 00 1 0 0.00824  10 1 1 0 0.00024  2 0 2 22  0.00281  0 00 2 0 0.00293  10 1 1 1 0.00806  2 111 1 0.00110  0 01 1 1 0.00012  1 1 10 0 0.00012  2 12 22  0.00110  0 02 2 2 0.00006  1 1 10 1 0.00793  2 2 111  0.00012  0 10 0 0 0.00824  1 1 1 0 2 0.00012  2 2 2 00  0.00006  0 10 0 1 0.00006  1 1 1 1 0 0.00793  2 2 2 01 0.00012  0 10 0 2 0.00012  1 1 1 1 1 0.38757  2 2 2 02  0.00269  0 11 0 1 0.00024  1 1 1 1 2 0.00110  2 2 2 10  0.00012  0 11 1 1 0.00806  1 1 12 0 0.00012  2 2 2 12  0.00110  0 20 0 0 0.00293  1 1 12 1 0.00110  2 2 2 20  0.00269  0 20 0 1 0.00012  1 12 2 2 0.00012  2 2 2 21 0.00110  0 20 0 2 0.00018  12 1 1 1 0.00110  2 2 2 22  0.07532  0 22 2 2 0.00281  12 2 2 2 0.00110  Total  1.00000  7  frequency  Note that Table 1.3 has fewer than 243 lines. It is because some patterns do not appear in the image. The high frequency of the patterns 00000, 11111, and 22222 reflects the fact that cross-shaped neighbourhoods of 5 pixels typically have the same colour. In general, the pattern distribution is unknown, and Meloche and Zamar (1994) propose to estimate it directly from the records  Y\,..., Y , n  for binary (black/white) images.  In this thesis, we extend this to the case of images with an arbitrary, butfinite,number of grey levels, and we develop a family of estimates of the pattern distribution. In the case of single-site neighbourhoods, the above frequency distribution can be written as  <?i(*) =,=-x; i i{*=*},  (2)  n  where l{z,- = 6} = 1 when the true colour at site i is equal to <5, otherwise, it is equal to zero. The subscript 1 is used to indicate that we are dealing with single-site neighbourhoods. More generally, for an arbitrary neighbourhood size r, the above frequency distribution can be written as  Qr(6)  = "i=i£ !Ki = • • •. = M, n  *i,T  (3)  and the 6 in equation (3) is r-dimensional. The subscript r in the above equation is used to indicate that we are dealing with r-site neighbourhoods. For binary images, Meloche and Zamar (1994) propose to use  where Q defined by equation (3) depends on the Zi, z ,.. •, z . Note that z,- depends r  2  8  n  on the typically unknown parameters Q and a. For this reason, Meloche and Zamar T  (1994) propose methods to estimate Q and cr in the case of C = {0,1}, that is, binary T  images. While equation (4) defines an estimate no matter what the colour set C is, the estimates of Q and a proposed by Meloche and Zamar (1994) are only valid in the case T  of C = {0,1}. In this thesis, we propose a family of simple estimates for Q and a in T  the case of C = {ci, c ,..., cjt}, that is, images with an arbitrary, butfinite,number of 2  grey levels. Chapter 2 will focus on the estimation of Q , assuming a is known, while r  Chapter 3 will deal with the joint estimation of Q and cr. r  9  2 Estimations of Q. 2.1 Introduction  Firstly, let Q , which will be used frequently, be the |C| vector of all possible r  r  Q (6ys, r  where |C| is the total number of patterns. Meloche and Zamar (1994) propose a biased r  estimate of Q based on indicator functions for binary images, and a general formula for r  obtaining an unbiased estimate from the biased one. In this chapter, we extend their idea to images with an arbitrary, but finite, number of grey levels. And we also discover a simple representation which makes the computations of the estimates easier. In this chapter we assume that a is known. The joint estimation of Q and a will be discussed r  in Chapter 3. 2.2 Unbiased Estimations of Q  r  We start by reviewing the biased and unbiased estimates of Q based on indicator r  functions for binary images proposed by Meloche and Zamar (1994). Later on, we will extend to image with more than two grey levels. Consider a case of 2 grey levels (eg. black/white), that is, C = {0,1}, with single-site neighbourhoods (r = 1). Let  Now, define <?i(0)  £ £ > ( o , t f )  n  . = i  10  1  n  1  and Qi to be the vector  (Qi(0), Qi(l)) . T  This Qi is a biased estimate of Qi because, for  example, EQ {0) l  =  £±E1{*;-<  n  I  = -EPf>,- + ere,- < -) n 2 = ^E{P( ,- < ±-)\{zi e  =  Similarly, EQ (l) x  =  IE1{Z,n  =  = 0}  + P(c-  < £)l{zi  = 1}}  0}$(i-) + ± E 1 { * = 1 } * ( ^ ) la n la  $(^-)<2i(0) + $(±)Qi(l),  where $(•) is the cumulative distribution  function of a N(0,1) random variable. Now we let a  =  b — $(57),  then we can  write the above in matrix notation. £Qi = AiQi where A =  tab  x  b a  So, an unbiased estimate can be obtained by inverting Ai and defining Qi = A Q , so 1  x  that EQi = A f ^ Q i = A^AiQi = Q . That is, x  and Qi is unbiased for Qi. 11  1  Using the same colour set C, we now consider linear 3-site neighbourhoods, that is, 3 pixels in a row, either vertically or horizontally, with C = {0,1}. As we have just mentioned in Example 1.1, there are 2 = 8 possible patterns, namely 000, 001, ..., 111. 3  Recall from equations (1) and (2) in Chapter 1 that 1 Qsi^l, <$2, fa) =  -  n  53 H i,l  ~ fa, i,2 = fa, Z,,3 = fa},  z  z  where <5-, = 0,1, i — 1,2,3. Following the same idea as above, define  0(000) =  ±Y:i{Yi,i<\}HY ,2<\}l{Y <\}  3  i  i>3  $3(001)  = I^i y;. <I}i y;, <I}i y;.3>I}  W i l )  =  {  1  {  2  {  ^HYi^lMY^^^Wt^^}.  Again, we write Q for the vector (g(000), $(001), $ (010),..., <2(111)). In the T  3  3  3  3  3  case of 2 grey levels, we use two indicator functions, \{Yi < |} for black pixel, and l{Yi > |} for white pixel (for single-site neighbourhoods). That is, if Y{ < | , then we say that the colour at site i is black. And if Yi > | then we say that the colour at site i is white. Observe that there is a relationship between Qi and Q . In the case of linear 3  3-site neighbourhoods, we use products of 3 indicators chosen from l{Yi < |}, l{Yi > |}. So Qi and Q have similar structure. Since Qi is biased for Qi, we would expect that 3  Q is also biased for Q3. 3  Now, we can compute the expectations of Q3(6i, 62, S )'s. For example, 3  EQ (000) 3  =  E^l{Y  itl  < \}\{Y  ia  < -}l{Y  12  l  i<3  < \)  = =  < \)P(Yia < \)P(Y  < i)  ii3  ±p{[P{ei  <  tl  =0}  + Pfa  = 0} + P  [ ^ , 3 < ^ ) l k 3  <  = 1}] x  ( e ^ < ^ ) l h 3 =  l}]}.  Performing similar calculations for the other seven expectations, and letting a be and 6 be $ ( 5 7 ) , we  ' 0(000)  get  2  ab  ab  ab  b  ab  ab  ab  ab  b  ab  <?(001)  ab  ab  b  ab  ab  Qs(OlO)  b  ab  ab  ab  Q (0ll)  a  a?b a?b ab  Q (ioo)  2  2  2  Qs(OlO)  a6  ab  a  <3(011)  a6  ab  ab  a  Q (ioo)  a6  ab  ab  b  g (ioi)  ab  ab  b  ab  ab  a  Q (no)  ab  b  ab  ab  ab  6 3  ab  ab  ab  ab  2  2  3  2  3  2  3  ^ $3(111),  v  2  2  2  2  3  2  3  2  3  3  3  2  2  2  3  3  >  3  2  2  2  3  2  2  2  2  3  2  2  2  2  3  2  2  2  3  3  >  ab  2  a6 a 2  Q(000)  (  ab  <? (ooi) 3  E  a6  3  >  3  2  3  2  2  3  2  2  ab  ab  <? (ioi)  ab  a  ab  Q (no)  ab  ab  3  2  2  2  2  3  2  3  2  3  2  2  3  , 03(111)  J  In matrix notation, we can write the above as £Q = A Q 3  3  3  As before, we obtain an unbiased estimate of Q by defining Q = A Q . 1  3  3  3  3  As shown in Meloche and Zamar (1994), the above matrices Ai, A 3 (as well as the corresponding ones for all r) can be analytically inverted. Meloche and Zamar (1994) show that, in general, EQ — A Q with T  r  r  A (6\7)H«) R  # (  *  13  , 7  W~  # (  *  , 7 )  ,  (5)  where A  r  is the ( £ , 7 ) entry of A , and  (£,7)  r  #(£,7) =  9  S = 7*:) is the number k  of  k such that 6 = "fk, for k = 1,2,..., r, and that k  A - (^,7) = (-1) 1  (6)  (6 - a ) 2  r  2  r  where a = $ ( 5 7 ) , 6 = $ ( 2 7 ) - Equation (6) is valid for all r-site neighbourhoods, where r  is arbitrary andfinite,with C = {0,1}. As an application of equations (5) and (6) in the case of 2-site neighbourhoods, an  unbiased estimate of Q2 is f  $2(00)  (  N  ab ba u\  Q (01)  aa ab aa bb ba  <? (10)  ba bb aa ab  2  2  ^0  2  (H) j  K  bb  -1  ba ab aa j  ' g(oo) ^ 2  Q(oi) 2  $2(10) ^ $2(11) j  where Q (00) = ^Wa  < \}W,2  2  < \}  g (oi) = -vm* < hma > h 2  n  1  &  We now extend this idea for estimating Q based on indicator functions to a colour set with more than 2 grey levels. We will notice from the following examples that the bias matrix A that needs to be inverted may not have a simple structure. r  Example 2.1 : Suppose C = {0,1,2} and r = 1. Now let <p(o, ) y  = i{y<^} 14  = ni<,<|) v(2,y)  =  H»>l)  and define n  n  t  e  n.  l  «=i  2  = 1  ",=i  2  2  Giro = ^Ev(2,«)=^i:i{«>|} t=i i=i ^ n  n  After carrying out some calculations, and letting a *(£) " = *(£) " * ( £ ) , / = 1 — $ ( ^ 7 ) , we obtain ' &(0) '  -  ^ 0l(2) ,  b  c^  d  e  f  g  h  i j  K  c = $(=jj-),d =  = 1" *(£)> * = 1 "  ' a  =  E  = $(57), b = $(37),  (  Qi(0)  =  y  Qi(i)  ^ Gi(2) j  or, in matrix notation, EQi = AiQx. When r = 2 (linear 2-site neighbourhoods), there are 3 = 9 possible patterns, namely, 00, 01, 02, 10, 11, 12, 20, 21, and 22. And again 2  by carrying out some calculations, we obtain  15  ' Q (oo)  ^ aa  ab ac  ba  bb  be ca  cb  cc)  g (oi)  ad ae af  bd  be  bf  cd  ce  cf  Q (oi)  <? (02)  ag ah  bg bh  bi  eg  ch  ci  Q (02)  <? (io)  da db dc  ea  eb  ec fa  fb  fc  Q (10)  dd de  df  ed  ee  ef  fd  fe  ff  0 (ii)  0 (12)  dg dh  di  eg  eh  ei  fg  fh  fi  Q (12)  0 (20)  ga gb gc  ha hb he  ia  ib  ic  Q (20)  <? (21)  gd ge 9f  hd he hf  id  ie  if  g (2i)  hg hh hi  ig  ih  ii j  N  2  2  2  2  E  =  Q (H) 2  2  2  2  ^ 0 (22) j 2  or EQ2  = A2Q2  ai  \ 99 gh  ' g (oo)  >  2  2  2  2  2  2  2  2  < S (22) j 2  in matrix notation: But now, A does not seem to have a simple 2  structure.* In the case of binary images, observe that the two bias matrices, Ai, A are closely 2  related. In fact A is a 2-fold Kronecker product of Ai. 2  Definition 2.1 (Kronecker product of matrices) : Let A = (a,j) be a p x m martix and B = (bap) be a q x n matrix. The pq x mn matrix with aijb p as the element in the i,ath row and the j,/?th column is called the a  Kronecker or direct product  of A and B and is denoted by A ® B; that is, ^ anB a B ... J2  A <g> B  a B im  a iB a B ... a B 2  22  y a iB a B p  p2  16  2m  ^  According to Anderson (1984), (A 0 B) = A 0 B , -1  - 1  (7)  - 1  that is, the inverse of a Kronecker product of 2 matrices is the Kronecker product of the inverses of the 2 matrices. In what follows, A^) denotes the r-fold Kronecker product r  of A, that is A  (r)  = A 0 A 0 • • • 0 A.  (8)  Since 6 and 7 are r-dimensional, we can write the (£,7) entry of A( ) as r  AV(6n)  = flA(6 , ).  »=i  (9)  i 7i  In case of binary images, we denote the matrix (bias part) of the 1-site case by Ai, the matrix of the 2-site case by A , and the matrix of the 3-site case by A 3 . Take a 2  closer look at Ai, and A . 2  A  a b 1  =  b a aa ab ba bb \  A = 2  ab aa bb ba  ' aAi  ba bb aa ab  ^ 6A1 aAi j  V bb  bAi  >  ba ab aa  By Definition 2.1, A is a 2-fold Kronecker product of Ai. Similarly, A 3 is a 3-fold 2  Kronecker product of Ai. Note that in Example 2.4, for the colour set of 3 grey levels, A is a 2-fold Kronecker 2  17  product of A i . Also note that A  in equation ( 6 ) is the (6,7) entry of the r-fold  (6,7)  1 T  Kronecker product of Aj" for the case of binary images. 1  Given a function <p : C x R —• R , define (  = - £ > ( * , * ) n  ,=1  Qr(6) = ^f:^(6 ,Y , ) p(8 ,Y , )...<p(6 ,Y , ) t=i 1  i 1  (  2  i 2  r  1 0  )  (11)  i r  n  In the examples so far, we have been using indicator function as the choice of <p(8, y). Proposition 2.1 :  If  ^,lM^, ^,2) • • • V(*r, X>)  Qr(8) = " E then  £Q =A* Q  (12)  r )  r  where  A\(8,7)  = Eip(8,~f  + ere), and A  ^  r  is the r-fold Kronecker product of A i .  A ^  is  a |C| x |C| matrix, where |C| is the size of the colour set. Furthermore, if |Ai| > 0, r  r  then Q = (Af ) ^ Q is an unbiased estimate of Q .» 1  r  r  r  Proof : We start the proof by expressing the equation EQ = A Q . We need to show r  .  EQ {6) = Y,M6,l)Qr{l) T  7  for all 6. Note that  18  r  r  (13)  EQ {6) T  =  E-J2v(^Y )--- (S Y ) iA  V  ri  itr  i^i  n  = - £ %>(*!, E 1{*M = 7l} • • • 7 1  i=l  K>) E 1{*> = 7r}  71  7r  1"  = -EEH^.l = 7l}^V(<5l,7l + ^0---E {^,r = -yr}E<p(8 ,*f 1  r  1=1 T i  n  1  n  = - t=l E E H i,l Z  "  r  + (Te ) T  7r  = 7l}  71  • • • E H^r  =  7r}^V(*l»7l + °"O • • • £V(<*r,7r e  +  7r  = X:^(7)Ai(cS ,70-"^i(^,70 a  7  = E U^i(A,7,)a(7) n  7  = E^(«,7)Qr(7) 7  = E4 («,7)Qr(7) r)  7  which is just equation (12). Therefore, £(Ai  r )  ) " Q r = ^ ( A r ^ W Q r = (A< ^ A ^ Q , r  1  =Q  r  ,  and Q = (A ) ^ Q is unbiased for Q . » r  a  r  x  r  Proposition 2.1 shows how to obtain an unbiased and consistent estimate of Q for r  any function  <p(6,y)  provided |Ai| ^ 0. We therefore have a family of unbiased and  consistent estimates. In order to avoid the problem of inverting a large matrix, we can invert Aifirstfor the case of single-site neighbourhood, then we can use Definition 2.1 to build the A; = (A" )^. 1  1  Example 2.2 : Suppose C = {0,1,2}, r = 3, and define ¥>(0,y) = 1 <p(hy) =  y  v>(2,y) = y  2  19  and as usual 1  ^  •=i For instance, 1  $3(0,0,0)  n  - V l - 1 - 1 = 1 Q (i,i,i)  1  3  $3(2,2,2)  Obviously, Q 3 is biased for Q  3  .  n  — £ *»M '  *«'.2 ' ^.3  Just notice, for instance, that  dently from the records. By definition, Ax ( 6«=1 ,7)  $3(0,0,0) = 1  7 t  = E(p(8,~f  + ere),  we have  Ax(0, ) = £ l = l 7  Ai(l,7) = J5(7 + ae) = 7 Ax(2, ) = £ ( 7 + at)  2  7  = 7 + cr 2  2  By Proposition 2.1, Q3 = (A_i)^C»3 is unbiased for Q , where 3  1  A  1  =  Ax(0,0)  Ax(0,l) Ax(0,2)  N  Ax(l,0) Ax(l,l) Ai(l,2) ^ ( 2 , 0 )  Ax(2,l) Ax(2,2)  1  0  2  J  and  I  1 _ 2i 2  =3 2  2  CT  2  - 1  -1 2  1/  1  Ax"^  1  2  —a 2  20  indepen-  Now, we express Q3 = (A_i)( )Q in matrix form. 3  3  '  $3(0,0,0)  t  >  (1 - £) (f) 2  2  $3(0,0,1)  ( l - - d ) V )  > V  $3(0,0,0)  /  2  ^  $3(0,0,1)  (i-^) (2) 2  $3(0,0,2)  $3(0,0,2)  \  2.3 A Simple Representation for Q  r  Proposition 2 . 1 provides a class of unbiased estimates Q of Q . The proposed estir  r  mates have the form Q = (A )( )Q , where -1  r  r  r  Qr(6) =  -J2^{6 ,Y ^{S ,Y )... {6 ,Y , ). i  i  2  ia  9  r  i r  ,=1  71  The matrix (A )( ) is the r-fold Kronecker product of Ai, and when we have a large -1  r  colour set, C, and a big r, (A )( ) is a large matrix. -1  r  In this section, we show that Q can be expressed as r  1  ^  Qr(6) = -n £  i=l  m,  >5.l)*(«2. Y*) • • • * ( « r . * > )  (14)  for some \&(t\ y)'s. Proposition 2.2 : If we define Qr(S)  =  -Ev&,YM6 ,Yu)---tp(6T,Y ,r), 3  i  then EQ = A( )Q , where A^^ is the r-fold Kronecker product of the bias matrix Ai, r  r  r  r  and by definition, Ai(£,7) =  Etp{6,i + at) 21  (Af is |C| x |C] ). If |Ai| > 0, then 0  r  r  Q = (A )^ ^Q 1  r  r  r  Q,  is unbiased and consistent for  7  and  r  ''  can be expressed as  Qr(6) = n  i=i  -it*(6uY , )---y(6 ,Y , ) i 1  r  i r  where  *(*,y)  = E^V(7,y).«  Proof:  7  ''  71  7r  = £•••E f t ••• 71  A  i\r ^ £  11  7r  V(TI ,  no • • • noi  1=1  = ^E{E---Eft^(7l,ni)-"^ >(7r,nr)]} U  1=1  71  r  7r  = ^E{(E^7^(7i ni))- -(E^^(7r,nr))} :  5  n  i=l  =  71  7r  -fyil{8 Y , )...^{8 ,Y )]. u  n  i i  T  i<r  i=i  N o t e that w i t h this representation, Q is easily seen to be unbiased. r  EV(6,-y + ae) =  ££Aj V(r, ) T  T =  E^fiT-^T.-y T  = 1{* = 7}. 22  7  Indeed,  And so 1^ EQr{6)  =  —E  D^(8i,  Yi i)^(8 , t  = -n1 EAH** = t=i i  =  Y ) • • • \P(cy, Y )  2  i>2  i>r  = s }--. i{  ^}i{^.,2  2  Zi<r  = ts> r  n  Qr(6).  Note that since Q (6) is an average of an r-dependent sequence, it converges to its mean. r  Therefore, Q {6) is an unbiased and consistent estimate of  Q (6).»  r  Example 2.2 (Continued) : Recall that  ip(0,y) =  r  l,c?(l,y) = y,ip(2,y)  equation (17), f  ( 1  *(o,y)' * ( i , y )  "  =  2  2  1  CT  2  2  -<r»  < *( >y) > 2  \  -3 2  -1 2  2  1 2  ^  '¥>(0,y) ^  -1 1 2  v(i>y)  y ^ v(2,y);  Therefore, we can write mv)  = ( i - y ) i + (^)y + (|)y , a  ¥(0,y) = (cr )l4,(2)y + (-l)y , 2  2  (^H-Hy)y  *(i,y) =  + (;;)y, 2  and so 03(0,0,0)  1A = -E*(0,ni)*(0,na)*(0,^), »=i n  $3(0,0,1) = -E*(0,ni)*(0,na)*(l.^). n  i=l  23  = y. 2  By  $3(2,2,2) =  - £^(2,Y , )^!(2,Y )^(2,Y , ). y  i 1  ii2  i 3  As stated in Proposition 2.2, such a new representation can be obtain generally.* In summary, we have the following general procedure for obtaining unbiased and consistent estimates for Q : Let C = {ci, c ,..., c^}, and r > 0. Take any (p : C x r  2  R —• R, and then define A as the |C| x |C|(= k x k) bias matrix with elements As*, = E<p(6,i + ere).  If |A| > 0, then define  Qr(6) = if; *(*i,ni)*(fc.na) •••*(*, no where  *(W = £ ^ V ( W i  Then Q (S) is an unbiased and consistent estimate of T  Q (S). r  Example 2.3 : We use Figure 2.1 as the true and unobserved image, and Figure 2.2 as the noisy and observed image. The colour set C is {0,1,2}, r is the neighbourhood size, and cr = 0.50. The following tables shows how far away the biased Q is from the true r  Q , and how good and close the unbiased Q is to the true Q . r  r  24  r  Table 2.1: Pattern distribution with r = 1, <p(6,y)  <5  QiV)  Qi(S)  =  Indicator functions  Qi(S)  0  0.48340 0.47498 0.48505  1  0.42505 0.37836 0.42074  2  0.09155 0.14667 0.09421  Total 1.00000 1.00000 1.00000  Table 2.2: Pattern distribution with r 8  QM  =  Qi(8)  1, <p(6,y) Qi(S)  0  0.48340 0.37644 0.48306  1  0.42505 0.22966 0.42380  2  0.09155 0.39390 0.09314  Total 1.00000 1.00000 1.00000  25  —  Power functions  Table 2.3: Pattern distribution with r = 2, <p(8, y) =Indicator functions 6  Qi{6)  W )  0 0 0.47192 0.34601 0.46966 0 1 0.00842 0.11395 0.01315 0 2 0.00305 0.01501 0.00224 1 0 0.00842 0.11377 0.01274 1 1 0.41541 0.20496 0.40092 1 2 0.00122 0.05963 0.00709 2 0 0.00305 0.01520 0.00266 2 1 0.00122 0.05945 0.00667 2 2 0.08728 0.07202 0.08488 Total 1.00000 1.00000 1.00000  26  T a b l e 2.4: Pattern distribution w i t h r = 2,  6  For  W)  <p(8,y) = Power functions W)  0 0  0.47192  0.09996  0.46610  0 1  0.00842  0.06099  0.00452  0 2  0.00305  0.10460  0.00598  1 0  0.00842  0.06099  0.00480  1 1  0.41541  0.07745  0.42015  1 2  0.00122  0.12890  0.00000  2 0  0.00305  0.10460  0.00570  2 1  0.00122  0.12896  0.00000  2 2  0.08728  0.23355  0.09275  Total  1.00000  1.00000  1.00000  r = 3 (linear 3-site neighbourhood), the columns 0,3(6), Qs(S), and Qs(6) are dis-  played graphically as barplots i n Figures 2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3f. F i n a l l y , for  r = 5 (cross-shaped neighbourhood), the columns Qs(6), Qs(6), and Qs(6) are also displayed graphically as barplots i n Figures 2.4a, 2.4b, 2.4c, 2.4d, 2.4e, 2.4f. W e have seen that different neighbourhood shapes and sizes give different pattern distributions. T h e y also give different estimates  {z\, z ,..., z }, because as mentioned 2  n  before, the estimate of the colour at each site is a function of the data, Yi, available i n a neighbourhood of that site. T h e estimate requires estimates of pattern distributions,  proposed by Meloche and Z a m a r (1994)  Q (6), and a. Recall from equation (4) that r  . _SgM»(Y --g)qr(g) t  27  where  tf>  denotes the r-dimensional normal density with mean zero and covariance o~ I , 2  a  T  and 6 denotes the colour/pattern at the center of the neighbourhood. Figures 2.5a to C  2.5h are the restored images based on different neighbourhood shapes and sizes, and different estimations of the pattern distributions.*  28  3 Joint Estimations of Q and a. 3.1 Introduction Meloche and Zamar (1994) propose estimating z,- by Us MY,- - 6)Q (6) '  Z%  r  where <5 denotes the colour/pattern at the center of the neighbourhood, and (f>„ is the C  normal density with mean zero and variance cr . Furthermore, our proposed estimation 2  of Q involves cr, and we have been assuming that a is known. Equation (16) states that r  The subscript a indicates that Q {6) involves a. But a is likely to be unknown and must r  be estimated from the noisy image. Ideally, we want to estimate both Q and cr, but we r  will focus on the estimations of Q and cr. Once a is estimated, Q can also be estimated. x  r  For the rest of this thesis, we write Q instead of Qi. In this chapter, we propose some estimating equations for cr. 3.2 Estimation of a using Estimating Equation derived from Note that  n  ti  i=i  n  = =  lj^E{  + cre)  -TE(z i=i  + 2z ae + a e )  2  Zi  2  n  29  2  i  2  E(^Yli=i  S i=l ^6 Q(8). n  =  v+ 2  2  Thus, if Q<r is any unbiased estimate of Q, E(o-  2  - - E^ + E 2  t=l  = 0-  (18)  a  The subscript a is used to indicate that Q„ involves a. Now define  A„(a) = a - - X X + E " W) 2  ( )  6  19  and taking expection of equation (19), A(a) = EX (a) n  = a - (a 2  2  +£  6 Q(6)) 2  + £ eS £g (cT). 2  o  (20)  Note that according to equation (18), a is one of the root of A(a). Empirical evidence suggests that when Q is derived from indicator functions, A(a) has 2 roots, the smaller of which is a (irrespective of the colour set C). Theoretical results to that effect have not been reached yet. When Q,, is derived from power functions, and when C = {ci,c }, equation (19) 2  yields a closed-form estimate of a instead of a. 2  30  A(a) = a -(a + c g(c ) + c g(c )) + c g(c ) + c g(c ). 2  2  2  2  2  1  2  2  1  2  By solving A (a) = 0, tr can be estimated as 2  n  rai=l  C  -  2  Ci  This is just the estimate of a proposed by Meloche and Zamar (1994) for the colour set 2  {0,1}. It is unbiased and consistent. When |C| = 3, it can be shown that A(a) = 6 if i  is derived from power functions.  y) = y,<p(cs,y) i n  Q(ci)  =  Q(C2) =  1  n  i^i  n  1  Q{cz)  =  n  n  ,ti  then Q = A Q, where a  A  a  =  l  1  1  Cl  c  c  ^ c{ + a  2  4 +a  2  3  4 + °' )  2  and ' A7  J  =  (c - c )(a - c c )  (c - c )(c + c )  -(ci - c ) ( a - c c )  -(ci - c )(c + c )  (ci - c )(a - cic )  (ci - c )(ci + c )  2  2  3  2  3  2  2  |A  3  x  3  3  2  3  3  x  3  -(c -c ) \ >  2  3  (c - c ) x  3  a 2  v  2  2  2  31  2  -(ci - c ) ) 2  where | A| = —(ci  —  c )(ci — c3)(c2 2  — c ). We can write equation (20) in matrix form. Let 3  X(a) =  a -(a  =  a -cr  2  + Z6 Q(S)) +  2  2  ,  S  +  2  ^ 6 EQ (6)  2  2  t  a  6  '£6 EQ (6)-Y;S Q(S). 2  a  s  2  s  Therefore, A(o)  = a -a 2  =  2  + A A: A Q T  T  a  a - a + A (A: 2  + AQ  1  2  T  -I )Q,  A  1  3  a  where I  Vc C -ClC2+cJ-ClC3 2  v C2 C3-ClC2+cJ-ClC3  /  3  )  - ( C2 C3 - C l 3  2  /  a -<7*  /  \  ^  - C 1 C 3 + C 1 C 2 + C 3 - C 2 C 3  *  2  a -c/ 2  +cj - C l  - C i C + C lC2+C2C  -CiC3+ClC2+C C3-C^  — C l C 3+ C l C 2 + C 2 C 3 -C?, I  C2  3  C3  - C ^  2  V-ClC3+ClC2+ci-C C  -ClC3+ClC2+C?-C2C3  2  3  /  and -c (a - a ) 2  2  2  2  C C — C C + q — CiC 2  3  a  c\{a - a )  2  3  + —CiC  3  2  + CiC + c c — c | 2  2  3  -clo* -* ) 2  2  -C1C3 + cic + ci - C C 2  = -(a 2  2  3  a ). 2  As a result, A(a) = a - a - (a 2  2  2  a ) = 0. 2  Thus, when |C| = 3, A(a) defined by equation (20) is identically zero. To get an estimate of a when |C| = 3, one possibility is to derive a different estimating equation starting 2  from higher moments of Y{. For example, we start from Y*. 32  3.3 Estimation of a using Estimating Equation derived from E(^YZ=\ Y ) 4  By simple computation, = 3cr + 6<T E 6 Q(S) 4  E{-IZY ) 4  2  (21)  + X 6 Q(6).  2  4  s  S  Th  Thus, if Qo- is any unbiased estimate Q , E(3<r + 4  6<r J2 S Q {6) 2  2  a  + £  S  -I£Y )  6 Q (8) 4  4  a  t  .8  7  1  =  0.  (22)  t'=l  Define 3a + 6a £ * < ? a ( < ! > ) + £ tf ^*) - - £ 1? and taking expectation of equation (23), 2 A(a) = 3a + 6a £ * £0„(*) + £ ^ . ( t f ) - £( - £ 1?). X (a) =  4  2  (23)  4  2  n  4  (24)  2  6  6  n  i=l  According to equation (22), a is one of the root of A(a). By equation (14),  D(c - )(a 2  Oaiei) =  &(<*)  2  =  C3  c ) + (4 - 4)Yi -  -  2C3  T ^ D C c i - ^ - C ^ ^  l l A  n  , = l  Then note that A (a) and A(a) defined by equations (23) and (24) are second degree n  polynomials in a . As a result, A(a) has 2 roots. The smaller of which is a . In general, 2  2  a can be estimated as the smallest root of A„(a). In particular, when C = {0,1,2}, 2  $•(0) =  ^tli  -  1  + (T  ) Y i  +  (  \  A  ) Y  Qa(l) = iD(« )l + (2)15 + (-1)1?] 2  n  ,=1  4.(2) = ^ D ( ^ ) i + (y)^  33  +  (^] 2  By solving A (a) = 0, we have n  1  7  n  6  n «=i  N  a + Q(l) + 40(2) - - ± \/(Q(l) + 4Q(2) - ^) . 2  2  We can conclude that for the colour set {0,1,2}, cr can be estimated as 2  i  n  7  «=i  Although «T may not be unbiased, it is consistent. The estimate of a can be obtained 2  by taking the square root of a . In general, we can derive an estimating equation for cr 2  2  with an arbitrary colour set in a similar fashion when Q is derived from power functions. r  34  4 Neighbourhood Shapes.  We have seen in the previous examples that the bigger the neighbourhood size, the better the restoration performance. But this will no longer hold if the neighbourhood size is big while the image is small. Sometimes a good and small neighbourhood may result in better performance in restoration than a big and bad one. The performance of the estimates zi,...,z  can be measured by the average expected square error:  n  'AMSE  = - JT E(Bi - Zi)  2  n  ,=i  According to the theorem by Chan and Meloche (1995),  AMSE  <7(1 2  — o- I (r) 2  0  a  * Q))  = -n Y, E(z - zi) = <T (1 - o h(n 2  ,-=i  2  2  {  a  * Q)).  (25)  can be obtained by numerical integration or Montecarlo. Iofoo- * Q) is  the middle element on the diagonal of the Fisher information matrix oir] *Q (recall that a  our vectors are indexed from 1 to r, so the middle element on the diagonal element is the (^2 ) L  th  element. r\ is the normal density with mean zero and variance cr. Figure 3.1 is a 2  a  129 x 129 true image with alternating horizontal strips of 0, 1, and 2, and Figure 3.2 is the degraded version of the true image with a = 0.50. Table 3.1 provides the approximate AMSE  (obtained by Montecarlo) for various neighbourhood sizes and shapes. In the  table, AMSE^  =  <7(1 - <7 / (77a * Q)),  AMSE  =  <T (1 - a I {q  AMSE  =  CT (1 - (T I (Vc * Qpower)),  2  3  2  2  0  2  2  0  2  a  2  0  35  * Qindicator))  where Q is the true pattern distribution, Qindicator is the estimated pattern distribution based o n the indicator functions, a n d Q wer is the estimated pattern d i s t r i b u t i o n based p0  on the power functions.  T a b l e 4 . 1 : AMSE  for various neighbourhood sizes a n d shapes  neighbourhood  AMSEx  AMSE  linear 3-site nbhd  0.042  0.050  0.054  0.021  0.032  0.031  0.013  0.042  0.080  0.0053  0.043  0.056  0.0025  0.045  0.043  0.0042  0.092  0.125  AMSE  2  3  (horizontal) linear 3-site nbhd (vertical). linear 5-site nbhd (horizontal) linear 5-site n b h d (vertical) 5-site n b h d (cross-shaped) linear 7-site nbhd (horizontal)  Note that when the Q (Sys r  are known, the cross-shaped neighbourhood results i n a  better performance w i t h the lowest AMSE\  than the other neighbourhoods used here.  In particular, i t performs better than a larger neighbourhood which is the linear a n d horizontal 7-site neighbourhood. In pratice, we substitute Q (6ys for Q r (5)'s i n equation T  (25) when we only have a noisy image. W e choose the neighbourhood shape which gives the smallest  AMSE. 36  When we use the estimates Qr(6~)'s in equation (25) for obtaining the AMSEi and AMSE3,  the linear and vertical 3-site neighbourhood seems to be the best choice be-  cause its corresponding $3(o")'s are relatively more accurate than the <5s(6)'s and <57(o")'s. When the neighbourhood size gets larger, the estimates Q 's become less accurate. Therer  fore, larger neighbourhood results in a worse restoration performance for our particular noisy image in Figure 3.2. If we have a larger image, the larger neighbourhoods may perform better than the smaller ones because the estimates Q for the larger neighbourhoods r  become more accurate.  37  5 Conclusions. By extending the idea of estimating Q  based on indicator functions by Meloche and  r  Zamar (1994), we have developed a family of estimates of Q  r  which can be based on any  arbitrary choice of a set of ip(6, y)'s. We start with any set of <p(6, y)'s which seems to have a simple structure. Then we define Qr(6)  = ^EniM*2,Ik)  • • • <p(6r,Y  )  itT  ,=1  n  which is biased for Q {6). By applying the propositions stated before, we obtain a new r  set of \&(c5, y)'s, where *(*,y) = £A£r¥>(7,v), 1  such that Qr(6) = - £  " ,=1  n0*(*2, ^,2) • • • * ( * r , K > )  is an unbiased and consistent estimate of Q (6). At the moment, we lack the theoretical r  results on judging which set of $(6, y)'s give the best and the most accurate estimate of Qr-  We have addressed the problem of estimating a by proposing some estimating equations for a. We have derived an estimating equation from E(^2~27=i Y?) f °  ra  colour set  with |C| = 2. By solving A„(a) = 0, we have obtained an estimate of a for any colour 2  set with |C| = 2. But the estimating equation derived from colour set when Q equation from  r  2~2?=i Y?) does not work for a bigger  is derived fron power functions. So we have derived another estimating Z)"=i Y*) f °  r a  colour set with |C| = 3. Again, by solving A (a) = 0, n  38  we have obtained an estimate of a for any colour set with | C | = 3. To estimate a 1  2  when | C | > 3, we can derive a different equation starting from a higher moment of Y{. In general, we estimate a by deriving an estimating equation for a in this fashion. The 2  2  esimate may not be unbiased, but it is consistent. By taking the square root of a , we 2  obtain the estimate of a. When Q is derived from indicator functions, the empirical evidence suggests that r  A(a) has 2 solutions for a," and the smaller of which is cr. Irrespective of the colour set C, estimating equation (18) always works. Therefore, we do not have to consider higher moments of Y,. But theoretical results have not been reached yet.  39  REFERENCES Anderson, T.W. (1984) An Introduction to Multivariate Statistical Analysis. Wiley.  Chan, A., Meloche, J. (1995) Estimation of a Gaussian Mean and Image Restoration. To be submitted for publication.  Meloche, J., Zamar, R. (1994) Black and White Image Restoration. Journal of Statistics. 22, 3, 335-355.  40  The Canadian  F i g u r e 2.2: Noisy Image (Observed)  ^  ?, s ! * y h '  '« - < j k $ M m * * s V - ^ •  41  Figure 2.3: Pattern distribution with r = 3, and (i) (p(8, y) = Indicator functions  (a) Q (S) vs 8 3  (b) Q {S) vs  8  (c) Qz{8) vs  8  Z  42  Figure 2.3: Pattern distribution with r = 3, and (ii) <p(6, y) = Power functions  SSS8888  m  wwww  Kawssg 888888  (e) Q (6) vs 3  43  6  Figure 2.4: Pattern distribution with r = 5, and (i) y>(6, y) = Indicator functions  (a) Q (S) vs 6 5  LL—tn.  & -—- »  , ...»X.« (b) &(<!>) vs £  •*—*  -  -»••  (c)  t  vs 6  44  Figure 2.4: Pattern distribution with r = 5, and (ii) ip(6,y) = Power functions  (d)Qs(6)vs6  (e) Q (S) vs 8 5  }  t  (f) Q (6) vs ,5 5  45  Figure 2.5: Restored Image, ip(6,y) = I n d i c a t o r  Figure 2.5: Restored Image, <p{6,y) = Power  F i g u r e 3.1: True Image (0-1-2 strips)  F i g u r e 3.2: Noisy Image (0-1-2 strips)  48  

Cite

Citation Scheme:

    

Usage Statistics

Country Views Downloads
United States 16 3
Japan 6 5
Philippines 5 0
Canada 4 0
Ukraine 4 0
China 4 6
Russia 3 0
India 2 0
Germany 1 19
Poland 1 0
City Views Downloads
Unknown 20 30
Tokyo 6 0
Atlanta 5 0
Shenzhen 4 1
Vancouver 3 0
Ashburn 3 0
Saint Petersburg 3 0
Mangalore 1 0
Berlin 1 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}
Download Stats

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.831.1-0086896/manifest

Comment

Related Items