Chapter 3

Text and image compression

1
3.2 Compression principles

Source encoders and destination

decoders
Lossless and lossy compression
Entropy encoding
Source encoding

2
3.2.1 Source encoder and destination
decoders
 Figure 3.1

3
3.2.2 Lossless and lossy compression

 Lossless compression algorithm, when the

compressed information is decompressed,
there is no loss of information to be
reversible
 Lossy compression algorithms, is normally
not to reproduce an exact copy of the source
information after decompression
 Example application of lossy compression
are for the transfer of digitized images and
audio and video streams
4
 3.2.3 Entropy encoding

 Run-length encoding
 When the source information comprises long
substrings of the same character or binary
digit.
 Insteadof transmitting these directly, they
are sent in the form of a string of codewords,
each indicating both the bit - 0 or 1 - and the
number of bits in the substring
0000000111111111110000011… 5
 3.2.3 Entropy encoding
 This could be represented as:0,7,1,10 0,5,1,2…
 Alternatively, if
we ensure the first substring
always comprises binary 0s, then the string
could be represented as 7,10,5,2…
 Statistical encoding
 Statisticalencoding exploits this property by
using a set of variable length codewords with
the shortest codewords used to represent the
most frequently occurring symbols
6
3.2.3 Entropy encoding

 Ensure that a shorter codeword in the set

does not form the start of a longer codeword
otherwise the decoder will interpret the
string on the wrong codeword boundaries
 prefix property
 Huffman coding algorithm
 Minimum average number of bits that are
required to transmit a particular source
stream is known as the entropy of the source
7
3.2.3 Entropy encoding
 Shannon:
n
Entropy, H    Pilog 2 Pi
i2

 Average number of bits per codeword :

n

N
i 1
i Pi

8
3.2.3 Entropy encoding
 Example 3.1
-- Comprises only the six different characters
M,F,Y,N,0 AND 1
-- Frequency of occurrence of 0.25, 0.25, 0.125,
0.125, 0.125, and 0.125
-- If the encoding algorithm under following
set of codewords:
M=10, F=11, Y=010, N=011, 0=000,1=001
9
3.2.4 Source encoding
 Differential encoding
-- Instead of using a set of relatively large
codewords to represent the amplitude of
each value/signal, a set of smaller codewords
can be used each of which indicates only the
difference in amplitude between the current
value/signal being encoded
-- For example, 12 bits to obtain the required
dynamic range but the maximum difference
in amplitude between successive samples of
the signal requires only 3-bits 10
3.2.4 Source encoding
 Transform encoding
-- As we scan across a set of pixel locations
-- The rate of change in magnitude will vary
from zero, if all the pixel values remain the
same
-- A high rate of change it each pixel magnitude
changes from one location to the next
-- The rate of change in magnitude as one
traverses the matrix gives rise to a term
known as spatial frequency 11
3.2.4 Source encoding
-- Figure 3.2

12
3.2.4 Source encoding
-- The human eye is less sensitive to the higher
spatial frequency components
-- If we can transform the original spatial
form of representation into an equivalent
representation involving spatial frequency
components, then we can more readily
identify and eliminate those higher
frequency components which the eye cannot
detect thereby reducing the volume of
information
13
3.2.4 Source encoding

-- The transformation of a two-dimensional

matrix of pixel values into an equivalent
matrix of spatial frequency components
discrete cosine transform(DCT)
- Figure 3.2(b)

14
3.3 Text compression

 Static coding
 Dynamic or adaptive coding

15
3.3.1 Static Huffman coding

 First analyzed and the character types and

their relative frequency determined
 Anexample of a Huffman code tree is
AAAABBCD
 A binary 0 for the left branch and a binary 1
for the right branch

16
3.3.1 Static Huffman coding
 Figure 3.3

17
3.3.1 Static Huffman coding
 listing
the resulting weights of all the leaf and
branch nodes in the tree starting with the smallest
weight and proceeding from left to right and from
bottom to top
4 × 1 + 2× 2 + 1×3 + 1×3 = 14 bits
 Example 3.2
 Figure 3.4
 Huffman codewords have the unique property that
a shorter codeword will never form the start of a
longer codeword prefix property
18
3.3.1 Static Huffman coding
 Figure 3.5

19
3.3.2 Dynamic Huffman coding

 The receiver being able to determine the

character that is received, can also carry out
the same modifications to its own copy of the
tree so that it can interpret the next
codeword received according to the new
updated tree structure
 Example : this is simple

20
3.3.2 Dynamic Huffman coding
 Figure 3.6

21
3.3.2 Dynamic Huffman coding

 Since three is empty, the uncompressed

ASCII char is sent and the tree is created.
(Fig 3.6 (a))
 The encoder first checks whether the
character is already present in the tree
 Ifit is not present, then the encoder sends
the current codeword for the empty leaf-
again determined by its position in the tree
- Fig. 3.6 (b), (c), (d), (e)
22
3.3.2 Dynamic Huffman coding
 The empty leaf being assigned to the 0-branch and
the character to the 1-branch
- Fig. 3.6 (b)
 Ifthe character is already present in the tree, then
the frequency of occurrence of the leaf node is
incremented by unity – Fig. 3.6 (g)(f)
 Ifit is, then the encoder sends the current
codeword for the character in the normal way
 First list the weights of the leaf and branch nodes
in the updated tree from left to right and from
bottom to top starting at the empty leaf
23
- Fig. 3.6 (a)(b)
 3.3.2 Dynamic Huffman coding

 Ifthey are all in weight order, all is well and the

tree is left unchanged
- Fig. 3.6 (a)(b)
 Ifthere is a node out of order, the structure of the
tree is modified by exchanging the position of this
node with the other node in the tree-to produce an
incremented weight order
- Fig. 3.6 ©(d)(e)(f)(g)

24
3.3.3 Arithmetic coding
 Arithmeticcoding yields a single codeword
for each encoded string of characters
 Thefirst step is to divide the numeric range
from 0 to 1 into a number of different
characters present in the message to be sent
 The size of each segment by the probability
of the related character
 Thefirst character to be encoded w is in the
range 0.8 To 0.9
25
3.3.3 Arithmetic coding
 Figure 3.7

26
 3.3.3 Arithmetic coding
 The segment for the character e, is from 0.8 to
0.83(0.8 + 0.3× 0.1)
 Character n from 0.83 to 0.86 (0.83 + 0.3× 0.1)
 The character e has a range from 0.8 to 0.809 (0.8+
0.3× 0.03), the character n from 0.809 to 0.818 (0.809
+ 0.3× 0.03)
 P136之解釋

 The maximum number of characters in a string is

determined by the precision with which floating-
point numbers
27
3.3.4 Lemple-Ziv coding
 A table containing all the possible character
strings
 The encoder sends only the index of where
the word is stored in the table and, on
receipt of each index, the decoder uses this
to access the corresponding word/string of
characters from the table
 Dictionary-based compression algorithm
 Example 3.3
28
3.3.5 Lemple-Ziv-Welsh coding
 Initially,
the dictionary held by both the
encoder and decoder contains only the
character set
 The remaining entries in the dictionary are
then built up dynamically by both the
encoder and decoder and contain the words
that occur in the text

29
3.3.5 Lemple-Ziv-Welsh coding
 Figure 3.8

30
3.3.5 Lemple-Ziv-Welsh coding
 The encoder prior to sending each word in
the form of single characters, first checks to
determine if the word is currently stored in
its dictionary and, if it is, it sends only the
index for the word
 Theavailable space become full, then the
number of entries is allowed to increase
incrementally

31
3.4 Image compression
3.4.1 Graphic interchange format(GIF)
 The resulting table of colors therefore
consists of 256 entries, each of which
contains a 24-bit color value
 Figure 3.9
 GIF also allow an image to be stored and
subsequently transferred over the network
in an interlaced mode
 The first contains 1/8 of the total
compressed image data, the second a further
1/8, the third a further 1/4 , and the last the
remaining 1/2 32
33
3.4.1 Graphic interchange format
 Figure 3.9

34
3.4.1 Graphic interchange format
 Figure 3.10

35
3.4.2 Tagged image file format

 The supports pixel resolutions of up to 48

bits-16 bits each for R, G, and B
 TheLZW compression algorithm that is
used

36
3.4.4 Digitized pictures
3.4.5 JPEG
-- There are five main stages associated with
this mode:image/block preparation, forward
DCT, quantization, entropy encoding, and
frame building

37
3.4.5 JPEG
-- Figure 3.14

38
3.4.5 JPEG

 Image/block preparation
-- Once the source image format has been
selected and prepared
-- Block preparation
-- It would be too time consuming to compute
the DCT of the total matrix in a single step
so each matrix is first divided into set of
smaller 8× 8 submatrics as a block

39
3.4.5 JPEG
-- Figure 3.15

40
 3.4.5 JPEG
 Forward DCT
-- Each pixel value is quantized using 8 bits
which produces a value in the range 0 to 255
for the intensity/luminance values-R,G,B or Y
- and a value in the range -128 to +127 for the
two chrominance values-Cb and Cr
-- All the values are first centered around zero
by subtracting 128
-- The input 2-D matrix is represented by :P[x,y]
and the transformed matrix by F[i, j]
41
 3.4.5 JPEG
 Figure 3.16

42
3.4.5 JPEG

 Quantization

-- The human eye responds primarily to the

DC coefficient and the lower spatial
frequency coefficients
-- Quantization phase by dropping-in practice,
setting to zero-those spatial frequency
coefficients in the transformed matrix whose
amplitudes are less than a defined threshold
value
43
3.4.5 JPEG
 The quantization process aims to reduce the
size of the DC and AC coefficients so that
less bandwidth is required
 A division
operation is performed using the
defined threshold value as the divisor
 Example 3.4
 The quantization table with the threshold
value to be used with a particular DCT
coefficient in the corresponding position in
the matrix
44
3.4.5 JPEG
 Figure 3.17

45
3.4.5 JPEG
 An example set of threshold values is given in the
quantization table , it can conclude a number of
points:
-- The computation of the quantized coefficients
involves rounding the quotients to the nearest
integer value
-- The threshold values used, increase in magnitude
with increasing spatial frequency
-- The DC coefficient in the transformed matrix is
largest
-- Many of the higher-frequency coefficients are zero
46
3.4.5 JPEG
 Entropy encoding
four steps:vectoring, differential encoding, run-
length encoding, and Huffman encoding
 Vectoring

-- represent the values in the form of a single-

dimension vector as vectoring
-- Zig-zag scan
-- the DC coefficient and lower-frequency AC
coefficients-both horizontal and vertical-are
scanned first
47
 3.4.5 JPEG
 Figure 3.18

48
3.4.5 JPEG
 Differential encoding
-- the first element in each transformed block
is the DC coefficient which is a measure of
the average color/luminance/chrominance
-- the DC coefficient varies only slowly from
one block to the next
-- differential encoding since this encodes only
the difference between each pair of values
-- for example, if the sequence of DC
coefficeints in consecutive quantized blocks-
one per block –was: 12, 13, 11, 11, 10, … 49
3.4.5 JPEG

-- the corresponding difference values would be : 12,

1, -2, 0, -1, …
-- the difference values are then encoded in the form
(SSS,value)
-- the SSS field indicates the number of bits needed
to encode the value
-- the value field the actual bits that represent the
value
-- Figure 3.19
-- Example 3.5 50
51
3.4.5 JPEG
 Run-length encoding
-- The remaining 63 values in the vector are
the AC coefficients
-- Each pair is made up of (skip, value) where
skip is the number of zeros in the run
-- Figure 3.18 would be encoded as:
(0,6) (0,7)(0,3)(0,3) (0,3) (0,2)(0,2)(0,2)(0,2)(0,0)
-- Final pair (0,0) indicates the end of the
string for this block and that all the
remaining coefficients in the block are zero 52
3.4.5 JPEG

 Huffman encoding

-- Significant levels of compression can be

obtained by replacing long string of
binary digits by a string of much
shorter codewords
-- The length of each codeword being a
function of its relative frequency of
occurrence

53
3.4.5 JPEG

 Frame building
-- The frame builder is to encapsulate all the
information relating to an encoded
image/picture in this format
-- Figure 3.20

54
3.4.5 JPEG
-- Figure 3.20

55
 3.4.5 JPEG
 Jpeg decoding
-- Figure 3.21

56
Exercises
 Section 3.2 -- 3.3
 Section 3.3 -- 3.8 3.11
 Section 3.4 -- 3.25 3.31

57