2. UNIT IV Image Compression

3. Outline Goal of Image Compression Lossless and lossy compression Types of redundancy General image compression model Transform coding JPEG and its different modes JBIG standard EZW and SPIHT coding

4. Goal of Image Compression The goal of image compression is to reduce the amount of data required to represent a digital image.

5. Approaches Lossless –Information preserving –Low compression ratios Lossy –Not information preserving –High compression ratios Trade-off: image quality vs compression ratio

6. Data ≠ Information Data and information are not synonymous terms! Data is the means by which information is conveyed. Data compression aims to reduce the amount of data required to represent a given quantity of information while preserving as much information as possible.

7. Data vs Information (cont’d) The same amount of information can be represented by various amount of data, e.g.: Your wife, Helen, will meet you at Logan Airport in Boston at 5 minutes past 6:00 pm tomorrow night Your wife will meet you at Logan Airport at 5 minutes past 6:00 pm tomorrow night Helen will meet you at Logan at 6:00 pm tomorrow night Ex1: Ex2: Ex3:

8. Definitions: Compression Ratio compression Compression ratio:

9. Definitions: Data Redundancy Relative data redundancy: Example:

10. Types of Data Redundancy (1) Coding Redundancy (2) Interpixel Redundancy (3) Psychovisual Redundancy Compression attempts to reduce one or more of these redundancy types.

11. Coding Redundancy Code: a list of symbols (letters, numbers, bits etc.) Code word: a sequence of symbols used to represent a piece of information or an event (e.g., gray levels). Code word length: number of symbols in each code word

12. Coding Redundancy (cont’d) N x M image r k : k-th gray level P(r k ): probability of r k l(r k ): # of bits for r k Expected value:

13. Coding Redundancy (con’d) Case 1: l(r k ) = constant length Example:

14. Coding Redundancy (cont’d) Case 2: l(r k ) = variable length variable length

15. Interpixel redundancy Interpixel redundancy implies that pixel values are correlated (i.e., a pixel value can be reasonably predicted by its neighbors). autocorrelation: f(x)=g(x) A A B B

16. Interpixel redundancy (cont’d) To reduce interpixel redundancy, the data must be transformed in another format (i.e., using a transformation) –e.g., thresholding, DFT, DWT, etc. Example: original thresholded (profile – line 100) threshold (1+10) bits/pair

17. Psychovisual redundancy The human eye does not respond with equal sensitivity to all visual information. It is more sensitive to the lower frequencies than to the higher frequencies in the visual spectrum. Idea: discard data that is perceptually insignificant!

18. Psychovisual redundancy (cont’d) 256 gray levels 16 gray levels/random noise 16 gray levels C=8/4 = 2:1 i.e., add to each pixel a small pseudo-random number prior to quantization Example: quantization

19. Measuring Information What is the minimum amount of data that is sufficient to describe completely an image without loss of information? How do we measure the information content of a message/image?

20. Modeling Information We assume that information generation is a probabilistic process. Idea: associate information with probability! Note: I(E)=0 when P(E)=1 A random event E with probability P(E) contains:

21. How much information does a pixel contain? Suppose that gray level values are generated by a random process, then r k contains: units of information!

22. Average information content of an image: units/pixel (e.g., bits/pixel) using How much information does an image contain? (assumes statistically independent random events) Entropy:

23. Redundancy: Redundancy (revisited) where: Note: if L avg = H, then R=0 (no redundancy)

24. Entropy Estimation It is not easy to estimate H reliably! image

25. Entropy Estimation (cont’d) First order estimate of H:

26. Estimating Entropy (cont’d) Second order estimate of H: –Use relative frequencies of pixel blocks : image

27. Estimating Entropy (cont’d) The first-order estimate provides only a lower- bound on the compression that can be achieved. Differences between higher-order estimates of entropy and the first-order estimate indicate the presence of interpixel redundancy! Need to apply some transformations to deal with interpixel redundancy!

28. Estimating Entropy (cont’d) Example: consider differences: 16

29. Estimating Entropy (cont’d) Entropy of difference image: An even better transformation could be found since: Better than before (i.e., H=1.81 for original image)

30. Image Compression Model

31. Image Compression Model (cont’d) Mapper: transforms input data in a way that facilitates reduction of interpixel redundancies.

32. Image Compression Model (cont’d) Quantizer: reduces the accuracy of the mapper’s output in accordance with some pre-established fidelity criteria.

33. Image Compression Model (cont’d) Symbol encoder: assigns the shortest code to the most frequently occurring output values.

34. Image Compression Models (cont’d) Inverse steps are performed. Note that quantization is irreversible in general.

35. Fidelity Criteria How close is to ? Criteria –Subjective: based on human observers –Objective: mathematically defined criteria

36. Subjective Fidelity Criteria

37. Objective Fidelity Criteria Root mean square error (RMS) Mean-square signal-to-noise ratio (SNR)

38. RMSE = 5.17RMSE = 15.67 RMSE = 14.17 Objective Fidelity Criteria (cont’d)

39. Lossless Compression

40. Lossless Methods - Taxonomy

41. Huffman Coding (coding redundancy) A variable-length coding technique. Symbols are encoded one at a time! –There is a one-to-one correspondence between source symbols and code words Optimal code (i.e., minimizes code word length per source symbol).

42. Huffman Coding (cont’d) Forward Pass 1. Sort probabilities per symbol 2. Combine the lowest two probabilities 3. Repeat Step2 until only two probabilities remain.

43. Huffman Coding (cont’d) Backward Pass Assign code symbols going backwards

44. Huffman Coding (cont’d) L avg assuming Huffman coding: L avg assuming binary codes:

45. Huffman Coding/Decoding Coding/decoding can be implemented using a look-up table. Decoding can be done unambiguously.

46. Arithmetic (or Range) Coding (coding redundancy) Instead of encoding source symbols one at a time, sequences of source symbols are encoded together. –There is no one-to-one correspondence between source symbols and code words. Slower than Huffman coding but typically achieves better compression.

47. Arithmetic Coding (cont’d) A sequence of source symbols is assigned to a sub- interval in [0,1) which corresponds to an arithmetic code, e.g., We start with the interval [0, 1) ; as the number of symbols in the message increases, the interval used to represent the message becomes smaller. α1 α2 α3 α3 α4α1 α2 α3 α3 α4 [0.06752, 0.0688)0.068 arithmetic code

48. Arithmetic Coding (cont’d) Encode message: α 1 α 2 α 3 α 3 α 4 0 1 1) Start with interval [0, 1) 2) Subdivide [0, 1) based on the probabilities of α i 3) Update interval by processing source symbols

49. Example [0.06752, 0.0688) or 0.068 (must be inside interval) Encode α1 α2 α3 α3 α4α1 α2 α3 α3 α4 0.2 0.4 0.8

50. Example (cont’d) The message is encoded using 3 decimal digits or 3/5 = 0.6 decimal digits per source symbol. The entropy of this message is: Note: finite precision arithmetic might cause problems due to truncations! - (3 x 0.2log 10 (0.2)+0.4log 10 (0.4))=0.5786 digits/symbol α1 α2 α3 α3 α4α1 α2 α3 α3 α4

51. 1.0 0.8 0.4 0.2 0.8 0.72 0.56 0.48 0.4 0.0 0.72 0.688 0.624 0.592 0.5856 0.5728 0.5664 0.5728 0.57152 0.56896 0.56768 0.56 0.5664 Decode 0.572 Arithmetic Decoding α1α1 α2α2 α3α3 α4α4 α3 α3 α1 α2 α4α3 α3 α1 α2 α4

52. LZW Coding (interpixel redundancy) Requires no priori knowledge of pixel probability distribution values. Assigns fixed length code words to variable length symbol sequences. Included in GIF, TIFF and PDF file formats

53. LZW Coding A codebook (or dictionary) needs to be constructed. Initially, the first 256 entries of the dictionary are assigned to the gray levels 0,1,2,..,255 (i.e., assuming 8 bits/pixel) Consider a 4x4, 8 bit image 39 39 126 126 Dictionary Location Entry 01.255 256- 511- Initial Dictionary

54. LZW Coding (cont’d) - Is 39 in the dictionary……..Yes - What about 39-39………….No * Add 39-39 at location 256 Dictionary Location Entry 01.255 256- 511- 39-39 As the encoder examines image pixels, gray level sequences (i.e., blocks) that are not in the dictionary are assigned to a new entry. 39 39 126 126

55. Example 39 39 126 126 Concatenated Sequence: CS = CR + P else: (1) Output D(CR) (2) Add CS to D (3) CR=P If CS is found: (1) No Output (2) CR=CS (CR) (P) CR = empty

56. Decoding LZW Use the dictionary for decoding the “encoded output” sequence. The dictionary need not be sent with the encoded output. Can be built on the “fly” by the decoder as it reads the received code words.

57. Run-length coding (RLC) (interpixel redundancy) Used to reduce the size of a repeating string of symbols (i.e., runs): 1 1 1 1 1 0 0 0 0 0 0 1  (1,5) (0, 6) (1, 1) a a a b b b b b b c c  (a,3) (b, 6) (c, 2) Encodes a run of symbols into two bytes: (symbol, count) Can compress any type of data but cannot achieve high compression ratios compared to other compression methods.

58. Combining Huffman Coding with Run-length Coding Assuming that a message has been encoded using Huffman coding, additional compression can be achieved using run-length coding. e.g., (0,1)(1,1)(0,1)(1,0)(0,2)(1,4)(0,2)

59. Bit-plane coding (interpixel redundancy) Process each bit plane individually. (1) Decompose an image into a series of binary images. (2) Compress each binary image (e.g., using run-length coding)

60. Lossy Methods - Taxonomy

61. Lossy Compression Transform the image into a domain where compression can be performed more efficiently (i.e., reduce interpixel redundancies). ~ (N/n) 2 subimages

62. Example: Fourier Transform The magnitude of the FT decreases, as u, v increase! K-1 K << N

63. Transform Selection T(u,v) can be computed using various transformations, for example: –DFT –DCT (Discrete Cosine Transform) –KLT (Karhunen-Loeve Transformation)

64. DCT (Discrete Cosine Transform) if u=0 if u>0 if v=0 if v>0 Forward: Inverse:

65. DCT (cont’d) Set of basis functions for a 4x4 image (i.e., cosines of different frequencies).

66. DCT (cont’d) DFTWHTDCT RMS error: 2.32 1.78 1.13 Using 8 x 8 subimages 64 coefficients per subimage 50% of the coefficients truncated

67. DCT (cont’d) DCT minimizes "blocking artifacts" (i.e., boundaries between subimages do not become very visible). DFT i.e., n-point periodicity gives rise to discontinuities! DCT i.e., 2n-point periodicity prevents discontinuities!

68. DCT (cont’d) Subimage size selection: 2 x 2 subimages original 4 x 4 subimages8 x 8 subimages

69. JPEG Compression Accepted as an international image compression standard in 1992. It uses DCT for handling interpixel redundancy. Modes of operation: (1) Sequential DCT-based encoding (2) Progressive DCT-based encoding (3) Lossless encoding (4) Hierarchical encoding

70. JPEG Compression (Sequential DCT-based encoding) Entropy encoder Entropy decoder

71. JPEG Steps 1.Divide the image into 8x8 subimages; For each subimage do: 2. Shift the gray-levels in the range [-128, 127] - DCT requires range be centered around 0 3. Apply DCT  64 coefficients 1 DC coefficient: F(0,0) 63 AC coefficients: F(u,v)

72. Example (non-centered spectrum) [-128, 127]

73. JPEG Steps 4. Quantize the coefficients (i.e., reduce the amplitude of coefficients that do not contribute a lot). Q(u,v): quantization table

74. Example Quantization Table Q[i][j]

75. Example (cont’d) Quantization

76. JPEG Steps (cont’d) 5. Order the coefficients using zig-zag ordering - Places non-zero coefficients first - Creates long runs of zeros (i.e., ideal for run-length encoding )

77. Example

78. JPEG Steps (cont’d) 6. Encode coefficients: 6.1 Form “intermediate” symbol sequence 6.2 Encode “intermediate” symbol sequences into a binary sequence.

79. Intermediate Symbol Sequence symbol_1 (SIZE) symbol_2 (AMPLITUDE) DC DC (6) (61) SIZE: # bits for encoding the amplitude of coefficient

80. DC Coefficients Encoding symbol_1 symbol_2 (SIZE) (AMPLITUDE) predictive coding:

81. Intermediate Symbol Sequence (cont’d) AC end of block symbol_1 (RUN-LENGTH, SIZE) symbol_2 (AMPLITUDE) SIZE: # bits for encoding the amplitude of coefficient RUN-LENGTH: run of zeros preceding coefficient If RUN-LENGTH > 15, use symbol (15,0), i.e., RUN-LENGTH=16 AC (0, 2) (-3)

82. Example: AC Coefficients Encoding Symbol_1 (Variable Length Code (VLC)) Symbol_2 (Variable Length Integer (VLI)) (1,4) (12)  (111110110 1100) VLC VLI # bits

83. Final Symbol Sequence

84. Effect of “Quality” (58k bytes) (21k bytes) (8k bytes) lower compression higher compression

85. Effect of “Quality” (cont’d)

86. Effect of Quantization: homogeneous 8 x 8 block

87. Effect of Quantization: homogeneous 8 x 8 block (cont’d) Quantized De-quantized

88. Effect of Quantization: homogeneous 8 x 8 block (cont’d) ReconstructedError Original

89. Effect of Quantization: non-homogeneous 8 x 8 block

90. Effect of Quantization: non-homogeneous 8 x 8 block (cont’d) QuantizedDe-quantized

91. Effect of Quantization: non-homogeneous 8 x 8 block (cont’d) Reconstructed Error Original:

92. WSQ Fingerprint Compression An image coding standard for digitized fingerprints employing the discrete wavelet transform (Wavelet/Scalar Quantization or WSQ). Developed and maintained by: –FBI –Los Alamos National Lab (LANL) –National Institute for Standards and Technology (NIST)

93. Memory Requirements FBI is digitizing fingerprints at 500 dots per inch with 8 bits of grayscale resolution. A single fingerprint card turns into about 10 MB of data! A sample fingerprint image 768 x 768 pixels =589,824 bytes

94. Preserving Fingerprint Details The "white" spots in the middle of the black ridges are sweat pores They’re admissible points of identification in court, as are the little black flesh ‘‘islands’’ in the grooves between the ridges These details are just a couple pixels wide!

95. What compression scheme should be used? Better use a lossless method to preserve every pixel perfectly. Unfortunately, in practice lossless methods haven’t done better than 2:1 on fingerprints! Does JPEG work well for fingerprint compression?

96. Results using JPEG compression file size 45853 bytes compression ratio: 12.9 The fine details are pretty much history, and the whole image has this artificial ‘‘blocky’’ pattern superimposed on it. The blocking artifacts affect the performance of manual or automated systems!

97. Results using WSQ compression file size 45621 bytes compression ratio: 12.9 The fine details are preserved better than they are with JPEG. NO blocking artifacts!

98. WSQ Algorithm

99. Varying compression ratio FBI’s target bit rate is around 0.75 bits per pixel (bpp) –i.e., corresponds to a target compression ratio of 10.7 (assuming 8-bit images) This target bit rate is set via a ‘‘knob’’ on the WSQ algorithm. –i.e., similar to the "quality" parameter in many JPEG implementations.

100. Varying compression ratio (cont’d) Original image 768 x 768 pixels (589824 bytes)

101. Varying compression ratio (cont’d) 0.9 bpp compression WSQ image, file size 47619 bytes, compression ratio 12.4 JPEG image, file size 49658 bytes, compression ratio 11.9

102. Varying compression ratio (cont’d) 0.75 bpp compression WSQ image, file size 39270 bytes compression ratio 15.0 JPEG image, file size 40780 bytes, compression ratio 14.5

103. Varying compression ratio (cont’d) 0.6 bpp compression WSQ image, file size 30987 bytes, compression ratio 19.0 JPEG image, file size 30081 bytes, compression ratio 19.6

104. JPEG Modes JPEG supports several different modes –Sequential Mode –Progressive Mode –Hierarchical Mode –Lossless Mode Sequential is the default mode –Each image component is encoded in a single left-to-right, top-to-bottom scan. –This is the mode we have just described!

105. Progressive JPEG The image is encoded in multiple scans, in order to produce a quick, rough decoded image when transmission time is long. Sequential Progressive

106. Progressive JPEG (cont’d) Each scan, encodes a subset of DCT coefficients. (1) Progressive spectral selection algorithm (2) Progressive successive approximation algorithm (3) Combined progressive algorithm

107. Progressive JPEG (cont’d) (1) Progressive spectral selection algorithm –Group DCT coefficients into several spectral bands –Send low-frequency DCT coefficients first –Send higher-frequency DCT coefficients next

108. Example

109. Progressive JPEG (cont’d) (2) Progressive successive approximation algorithm –Send all DCT coefficients but with lower precision. –Refine DCT coefficients in later scans.

110. Example

111. after 0.9safter 1.6s after 3.6safter 7.0s

112. Progressive JPEG (cont’d) (3) Combined progressive algorithm –Combines spectral selection and successive approximation.

113. Hierarchical JPEG Hierarchical mode encodes the image at different resolutions. Image is transmitted in multiple passes with increased resolution at each pass.

114. Hierarchical JPEG (cont’d) N x N N/2 x N/2 N/4 x N/4

115. MPEG VIDEO COMPRESSION

116. Embedded Zero Tree Wavelet Coding

117. Compare the two matrices 57670411521280134414721536 704640115610881344140815361600 76883212161472 15361600 832 96013441536 16001536 832 9601216153616001536 960896 10881600 1536 768 832 128014721600 448768704640128014081600 1212-306-146-54-24-68-404 30369028-208-4 -50-10-20-24072-16 8238-246848-64328 88-3216-48 -1616 20 -56-16 32-16 -88-480-16 44360880-16 0

118. A Multi-resolution Analysis Example Wavelet Transform

119. Discrete Wavelet Transform –Sub bands arise from separable application of filters LL 1 LH 1 HL 1 HH 1 First stage LH 1 HL 1 HH 1 Second stage LL 2 LH 2 HL 2 HH 2

120. Embedded Zero tree Wavelet algorithm (EZW) A simple, yet remarkable effective, image compression algorithm, having the property that the bits in the bit stream are generated in order of importance, giving a fully embedded (progressive) code. The compressed data stream can have any bit rate desired. Any bit rate is only possible if there is information loss somewhere so that the compressor is lossy. However, lossless compression is also possible with less spectacular results.

121. 120 EZW - observations 1.Natural images in general have a low pass spectrum, so the wavelet coefficients will, on average, be smaller in the higher subbands than in the lower subbands. This shows that progressive encoding is a very natural choice for compressing wavelet transformed images, since the higher subbands only add detail. 2. Large wavelet coefficients are more important than smaller wavelet coefficients. 631 544 86 10 -7 29 55 -54 730 655 -13 30 -12 44 41 32 19 23 37 17 -4 –13 -13 39 25 -49 32 -4 9 -23 -17 -35 32 -10 56 -22 -7 -25 40 -10 6 34 -44 4 13 -12 21 24 -12 -2 -8 -24 -42 9 -21 45 13 -3 -16 -15 31 -11 -10 -17 typical wavelet coefficients for a 8*8 block in a real image

122. Zero Tree Coding

123. Parent – Child relationship coefficients that are in the same spatial location consist of a quad-tree.

124. EZW Algorithm

125. EZW Algorithm contd..

127. Scanning order of sub bands

128. EZW - example

129. EZW – Example contd..

130. Contd..

133. 132 EZW - example

134. Set Partitioning in Hierarchical Trees (SPIHT) Algorithm

135. SPIHT

144. Bi-Level Image Compression

145. Page 144 Overview Introduction to Bi-Level Image Compression Existing Facsimile Standards:  G3 (MR)  G4 (MMR)  JBIG[1] New Bi-Level Standards:  JBIG2

146. Page 145 Definition: Bi-Level Multi-Level (Gray Scale) Bi-Level (Black & White)

147. Page 146 Properties of Bi-Level Images Mostly High Frequency Often Very High Resolutions:  Computer Monitor: 96dpi  Fax Machine: 200dpi 1 page fax (8.5” x 11” x 200dpi) ~=.5 Meg  Laser Printer: 600dpi (1 page = 4.2 Megs)  High-End Printing Press: 1600dpi (30 Megs!) Will often contain text, halftoned images and line-art (graphs, equations, logos, etc.)

148. Page 147 Existing Fax Standards T.4 (Group 3)  MH - Modified Huffman (and RLE)  MR - Modified Read Uses information from previous line Uses MH mode every k lines for error correction T.6 (Group 4)  MMR - Modified Modified Read Uses information from previous line Assumes Error-Free Environment

149. Page 148 Existing Fax Standards JBIG[1] (T.82 -- March, 1993) Joint Bi-Level Image Experts Group  Committee with Academic & Industrial members: ISO (International organization of National Bodies) ITU-T (Regulatory body of the United Nations) Arithmetic Coding (QM Coder) Context-based prediction Progressive Compression (Display)

150. Page 149 Existing Fax Standards Arithmetic Q Coder  Numerous variations: Q, QM, MQ Used by JBIG[1], JPEG, JBIG2 & J2K Different probability tables, byte markers, etc. Adaptive Coder 16-bit Precision (32-bit C register) Uses numerous Approximations:  Fixed Probability Table  No Multiplication

151. Page 150 New Standards JBIG2 (T.88 -- February 2000) First “lossy” bi-level standard Supports Three basic coding modes:  Generic (MMR or JBIG[1]-like arithmetic)  Halftone  Text Image can be segmented into regions  Each region can be coded with a different method

152. Page 151 JBIG2 - Generic Coding The core coding method of JBIG2 has not changed that much from previous methods There are two methods available in generic coding:  MMR (Group 4)  MQ Arithmetic Coding (similar to JBIG[1]) larger contexts are available: ? A AA A

153. Page 152 JBIG-2 Halftone Coding A halftone is coded as a multi-level image, along with a pattern and grid parameters The decoder constructs the halftone from the multi- level image and the pattern The multi-level image is coded as bi-level bit-planes, with the generic coder

154. Page 153 JBIG2 - Text Coding Each symbol is encoded in a dictionary with generic coding: And then, the image is constructed by adding images from the dictionary: yy xx The symbol ID and the (relative) co-ordinates are coded

155. Page 154 JBIG2 - Text Coding In actual documents, many symbols are very similar -- often due to scanning or spacial quantization errors Lossy Coding: Hard Pattern Matching Lossless Coding: Soft Pattern Matching

156. Page 155 JBIG2 - Soft Pattern Matching Soft Pattern Matching (refinement coding) is when a symbol is coded using a similar, previously coded symbol to provide additional context information. Already coded:To be coded: x?