Data Compression: The Complete Reference

More documents

Recommendations

Info

David SalomonData CompressionThe Complete ReferenceThird EditionWith 537 Figures, 6 in Full Color
Page 2 and 3: Data CompressionThird Edition
Page 6 and 7: David SalomonDepartment of Computer
Page 8 and 9: Preface to theThird EditionI was pl
Page 10 and 11: Preface to the Third EditionixAn N-
Page 12 and 13: Preface to theSecond EditionThis se
Page 14 and 15: Preface to the Second EditionxiiiDi
Page 16 and 17: Preface to theFirst EditionHistoric
Page 18 and 19: ContentsPreface to the Third Editio
Page 20 and 21: Contentsxix4.15 Adaptive Vector Qua
Page 22 and 23: IntroductionGiambattista della Port
Page 24 and 25: Introduction 3analysis of any compr
Page 26 and 27: Introduction 5Letter Freq. Prob. Le
Page 28 and 29: Introduction 7that 0.9n is an integ
Page 30 and 31: Introduction 9illuminated by consid
Page 32 and 33: Introduction 11per character)—the
Page 34 and 35: Introduction 13Description File nam
Page 36 and 37: 1Basic Techniques1.1 Intuitive Comp
Page 38 and 39: 1.1 Intuitive Compression 17The com
Page 40 and 41: 1.1 Intuitive Compression 19Letters
Page 42 and 43: 1.3 RLE Text Compression1.3 RLE Tex
Page 44 and 45: 1.3 RLE Text Compression 23StartCom
Page 46 and 47: 1.4 RLE Image Compression 25by much
Page 48 and 49: 1.4 RLE Image Compression 27is comp
Page 50 and 51: % Returns the run lengths of% a mat
Page 52 and 53: 1.4 RLE Image Compression 31Each pi
Page 54 and 55:
1.4.3 The BinHex 4.0 Format1.4 RLE
Page 56 and 57:
1.5 Move-to-Front Coding 35by the C
Page 58 and 59:
1.5 Move-to-Front Coding 37i Code S
Page 60 and 61:
1.6 Scalar Quantization1.6 Scalar Q
Page 62 and 63:
1.6 Scalar Quantization 41an image
Page 64 and 65:
2Statistical MethodsThe methods dis
Page 66 and 67:
2.1 Information Theory Concepts 45W
Page 68 and 69:
2.1 Information Theory Concepts 47A
Page 70 and 71:
2.1 Information Theory Concepts 49A
Page 72 and 73:
2.3 Prefix Codes 51ThepropertyofCod
Page 74 and 75:
2.3 Prefix Codes 532. The triplet (
Page 76 and 77:
2.3 Prefix Codes 55so it is ideal f
Page 78 and 79:
2.4 The Golomb Code 57anyadjacent1
Page 80 and 81:
2.4 The Golomb Code 59code of the n
Page 82 and 83:
2.4 The Golomb Code 61increase slow
Page 84 and 85:
2.4 The Golomb Code 63(for run and
Page 86 and 87:
2.5 The Kraft-MacMillan Inequality
Page 88 and 89:
2.7 Shannon-Fano Coding 67and a tot
Page 90 and 91:
2.8 Huffman Coding 69assign a bit o
Page 92 and 93:
2.8 Huffman Coding 71⋄ Exercise 2
Page 94 and 95:
2.8 Huffman Coding 73Starts0,50% 1,
Page 96 and 97:
2.8 Huffman Coding 75A 0.55B 0.25C
Page 98 and 99:
2.8 Huffman Coding 771. This code i
Page 100 and 101:
2.8 Huffman Coding 79p 5 =(a + b) +
Page 102 and 103:
2.8 Huffman Coding 81of documents a
Page 104 and 105:
2.8 Huffman Coding 831 2 3 4 5 6 7
Page 106 and 107:
2.9 Adaptive Huffman Coding 85escap
Page 108 and 109:
2.9 Adaptive Huffman Coding 87(16)(
Page 110 and 111:
2.9 Adaptive Huffman Coding 892.9.5
Page 112 and 113:
2.10 MNP5 91The first stage has bee
Page 114 and 115:
2.10 MNP5 93F PQ CF PQ CF PQ C00000
Page 116 and 117:
2.11 MNP7 95(i) After some more swa
Page 118 and 119:
2.12 Reliability 97People who use H
Page 120 and 121:
2.13 Facsimile Compression2.13 Facs
Page 122 and 123:
2.13 Facsimile Compression 1012. A
Page 124 and 125:
2.13 Facsimile Compression 103conse
Page 126 and 127:
2.13 Facsimile Compression 105(a)b
Page 128 and 129:
2.14 Arithmetic Coding 1070 0 0 0 0
Page 130 and 131:
2.14 Arithmetic Coding 109The outpu
Page 132 and 133:
2.14 Arithmetic Coding 111where Ran
Page 134 and 135:
2.14 Arithmetic Coding 113Char. Cod
Page 136 and 137:
2.14 Arithmetic Coding 115add some
Page 138 and 139:
2.14 Arithmetic Coding 117Low = 0 +
Page 140 and 141:
2.14 Arithmetic Coding 119three exa
Page 142 and 143:
2.15 Adaptive Arithmetic Coding 121
Page 144 and 145:
2.15 Adaptive Arithmetic Coding 123
Page 146 and 147:
2.16 The QM Coder 1251 1 0.75 0.72A
Page 148 and 149:
2.16 The QM Coder 127How can we use
Page 150 and 151:
2.16 The QM Coder 129Symbol C A Ren
Page 152 and 153:
2.16 The QM Coder 131Qe Hex Dec Dec
Page 154 and 155:
2.17 Text Compression 133P =0.5 weg
Page 156 and 157:
2.18 PPM 135probability.) Another r
Page 158 and 159:
2.18 PPM 137I pounded the keys so h
Page 160 and 161:
2.18 PPM 139Table 2.71 shows contex
Page 162 and 163:
2.18 PPM 141Order 2 Order 1 Order 0
Page 164 and 165:
2.18 PPM 143In Case 2, the next sym
Page 166 and 167:
2.18 PPM 145Figure 2.73 shows how s
Page 168 and 169:
2.18 PPM 147a,1 a,1 s,1 a,1 s,2a,2
Page 170 and 171:
2.18 PPM 149a,3 i,2 m,1 n,1 s,6n,1i
Page 172 and 173:
2.18 PPM 151Based on experience wit
Page 174 and 175:
2.18 PPM 153This complex procedure
Page 176 and 177:
2.18 PPM 155Order Context Symbol Co
Page 178 and 179:
2.19 Context-Tree Weighting 157θ:
Page 180 and 181:
2.19 Context-Tree Weighting 159Let
Page 182 and 183:
2.19 Context-Tree Weighting 1611b 3
Page 184 and 185:
2.19.1 CTW for Text Compression2.19
Page 186 and 187:
3Dictionary MethodsStatistical comp
Page 188 and 189:
3.1 String Compression 167data, use
Page 190 and 191:
3.3 LZ77 (Sliding Window) 169so it
Page 192 and 193:
3.3 LZ77 (Sliding Window) 171⌈log
Page 194 and 195:
3.4 LZSS3.4 LZSS 173This version of
Page 196 and 197:
3.4 LZSS 175The tree should be upda
Page 198 and 199:
3.5 Repetition Times 177There is no
Page 200 and 201:
3.6 QIC-122 179BytesLength2 003 014
Page 202 and 203:
3.7 LZX 181trees (Section 2.8.6) to
Page 204 and 205:
3.8 File Differencing: VCDIFF3.8 Fi
Page 206 and 207:
3.9 LZ78 185The number of possible
Page 208 and 209:
3.9 LZ78 187null4-␣8-a22-c 6-d 16
Page 210 and 211:
3.10 LZFG 189is written on the comp
Page 212 and 213:
3.11 LZRW1 191ilar to A1 and A2 but
Page 214 and 215:
3.11 LZRW1 193random pointer4096poi
Page 216 and 217:
3.13 LZW 195partition (32 pointers)
Page 218 and 219:
3.13 LZW 197in new in newI dict? en
Page 220 and 221:
3.13 LZW 199are (1) it outputs the
Page 222 and 223:
3.13 LZW 201keep the example simple
Page 224 and 225:
3.13 LZW 20326526626726826927027127
Page 226 and 227:
3.13 LZW 2053.13.4 DifferencingThe
Page 228 and 229:
3.14 LZMW 207Out- Add toStep Input
Page 230 and 231:
3.16 LZY 209structure, which elimin
Page 232 and 233:
3.17 LZP 211Step Input Add to dict.
Page 234 and 235:
3.17 LZP 213S (the f) is written on
Page 236 and 237:
3.17 LZP 2157. The current symbol i
Page 238 and 239:
3.17 LZP 217Length Code Length Code
Page 240 and 241:
3.18 Repetition Finder 219according
Page 242 and 243:
3.19 UNIX Compression 221i α Q 0 (
Page 244 and 245:
3.21 The V.42bis Protocol3.21 The V
Page 246 and 247:
3.23 Deflate: Zip and Gzip 225Phill
Page 248 and 249:
3.23 Deflate: Zip and Gzip 227press
Page 250 and 251:
3.23 Deflate: Zip and Gzip 229dista
Page 252 and 253:
l_count[0] = 0;for (bits = 1; bits
Page 254 and 255:
3.23 Deflate: Zip and Gzip 233to ha
Page 256 and 257:
3.23 Deflate: Zip and Gzip 235the c
Page 258 and 259:
3.24 PNG 237The four critical chunk
Page 260 and 261:
3.24 PNG 239The heuristic recommend
Page 262 and 263:
3.25 XML Compression: XMill 241revi
Page 264 and 265:
3.27 CRC 243algorithm based on the
Page 266 and 267:
3.27 CRC 245definition). In our exa
Page 268 and 269:
3.29 Data Compression Patents 247us
Page 270 and 271:
3.30 A Unification 249where it will
Page 272 and 273:
4Image CompressionThe first part of
Page 274 and 275:
4.1 Introduction4.1 Introduction 25
Page 276 and 277:
4.1 Introduction 255CompressedPixel
Page 278 and 279:
4.1 Introduction 257principle they
Page 280 and 281:
4.2 Approaches to Image Compression
Page 282 and 283:
Page 284 and 285:
Page 286 and 287:
Page 288 and 289:
Page 290 and 291:
Page 292 and 293:
Page 294 and 295:
4.3 Intuitive Methods 273is a measu
Page 296 and 297:
4.4 Image Transforms 275the image t
Page 298 and 299:
4.4 Image Transforms 27710010090908
Page 300 and 301:
4.5 Orthogonal Transforms 279each p
Page 302 and 303:
4.5 Orthogonal Transforms 281To ill
Page 304 and 305:
4.5 Orthogonal Transforms 283values
Page 306 and 307:
4.5 Orthogonal Transforms 285M=3; N
Page 308 and 309:
4.5 Orthogonal Transforms 287Figure
Page 310 and 311:
4.6 The Discrete Cosine Transform 2
Page 312 and 313:
Page 314 and 315:
Page 316 and 317:
Page 318 and 319:
Page 320 and 321:
Page 322 and 323:
Page 324 and 325:
Page 326 and 327:
Page 328 and 329:
Page 330 and 331:
LLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLL
Page 332 and 333:
Page 334 and 335:
Page 336 and 337:
Page 338 and 339:
Page 340 and 341:
Page 342 and 343:
Page 344 and 345:
Page 346 and 347:
4.7 Test Images 325With the aid of
Page 348 and 349:
4.7 Test Images 327Figure 4.58: Man
Page 350 and 351:
4.8 JPEG4.8 JPEG 329JPEG is a sophi
Page 352 and 353:
4.8 JPEG 331no tables (or just a fe
Page 354 and 355:
4.8 JPEG 333The main international
Page 356 and 357:
4.8 JPEG 335The JPEG decoder works
Page 358 and 359:
4.8 JPEG 337If the two components o
Page 360 and 361:
4.8 JPEG 339Z0123456789ABCDEFR1 2 3
Page 362 and 363:
4.8 JPEG 341of Table 4.70 is 01, so
Page 364 and 365:
4.8 JPEG 343data units are combined
Page 366 and 367:
4.8 JPEG 345The JFIF marker (called
Page 368 and 369:
4.9 JPEG-LS 347and this is done wit
Page 370 and 371:
4.9 JPEG-LS 349In practice, we alwa
Page 372 and 373:
4.9 JPEG-LS 351the run). The two ma
Page 374 and 375:
4.10 Progressive Image Compression
Page 376 and 377:
Page 378 and 379:
Page 380 and 381:
Page 382 and 383:
4.11 JBIG 361each individually, as
Page 384 and 385:
4.11 JBIG 363O O OO O O O AO O ?(a)
Page 386 and 387:
4.11 JBIG 36501234567891011layer 11
Page 388 and 389:
4.11 JBIG 367The 10 periodic patter
Page 390 and 391:
4.12 JBIG2 369be compressed, since
Page 392 and 393:
4.12 JBIG2 371document. The encoder
Page 394 and 395:
4.12 JBIG2 3733. A procedure to dec
Page 396 and 397:
4.12 JBIG2 3751 1 0 0 01 1 1 0 0 0
Page 398 and 399:
4.12 JBIG2 377•• •Figure 4.10
Page 400 and 401:
4.12 JBIG2 379One page of the docum
Page 402 and 403:
4.13 Simple Images: EIDAC 381actual
Page 404 and 405:
4.14 Vector Quantization 383The pro
Page 406 and 407:
4.14 Vector Quantization 385how the
Page 408 and 409:
4.14 Vector Quantization 3872402202
Page 410 and 411:
4.14 Vector Quantization 389perturb
Page 412 and 413:
4.15 Adaptive Vector Quantization 3
Page 414 and 415:
4.15 Adaptive Vector Quantization 3
Page 416 and 417:
4.16 Block Matching 395012.0 1 2 ..
Page 418 and 419:
4.16 Block Matching 397Figure 4.117
Page 420 and 421:
4.17 Block Truncation Coding 399of
Page 422 and 423:
4.17 Block Truncation Coding 401As
Page 424 and 425:
and its preservation gives rise to
Page 426 and 427:
4.18 Context-Based Methods 405Compr
Page 428 and 429:
4.18 Context-Based Methods 40700000
Page 430 and 431:
4.19 FELICS 409A BPA B PBA PProbabi
Page 432 and 433:
4.20 Progressive FELICS4.20 Progres
Page 434 and 435:
4.20 Progressive FELICS 413(a)(b)Co
Page 436 and 437:
4.21 MLP 415If the value of the pix
Page 438 and 439:
4.21 MLP 417Table 4.133 shows the 1
Page 440 and 441:
4.21 MLP 419number xof errors xxx(a
Page 442 and 443:
4.21 MLP 4213.2. While the pixels o
Page 444 and 445:
4.22 Adaptive Golomb 423Variance Va
Page 446 and 447:
4.23 PPPM 425for text, where we can
Page 448 and 449:
4.24 CALIC 4270,0 0,1 0,2 0,3 0,4 0
Page 450 and 451:
4.25 Differential Lossless Compress
Page 452 and 453:
4.26 DPCM 431“The nerve impulse,
Page 454 and 455:
4.26 DPCM 433how the current data i
Page 456 and 457:
The total square prediction error i
Page 458 and 459:
4.28 Block Decomposition 437An alte
Page 460 and 461:
4.28 Block Decomposition 439AxPB(5,
Page 462 and 463:
4.29 Binary Tree Predictive Coding
Page 464 and 465:
Page 466 and 467:
Page 468 and 469:
Page 470 and 471:
4.30 Quadtrees 449subquadrants that
Page 472 and 473:
4.30 Quadtrees 45100 01 02 03 10 11
Page 474 and 475:
4.30 Quadtrees 453children, which a
Page 476 and 477:
4.30 Quadtrees 455pixels becomes a
Page 478 and 479:
4.30 Quadtrees 457Diff.Huffmanvalue
Page 480 and 481:
4.30 Quadtrees 459One advantage of
Page 482 and 483:
4.30 Quadtrees 461is indicated by a
Page 484 and 485:
}else{ // knot->depth == Log2(N) e
Page 486 and 487:
4.31 Quadrisection 465and set the d
Page 488 and 489:
Equation (4.45) is I =(0, 0, 0, 3,
Page 490 and 491:
4.31 Quadrisection 469by ignoring r
Page 492 and 493:
4.31 Quadrisection 471The two halve
Page 494 and 495:
4.32 Space-Filling Curves4.32 Space
Page 496 and 497:
4.33 Hilbert Scan and VQ 475close o
Page 498 and 499:
4.34 Finite Automata Methods 477the
Page 500 and 501:
4.34 Finite Automata Methods 479We
Page 502 and 503:
4.34 Finite Automata Methods 4810 (
Page 504 and 505:
4.34 Finite Automata Methods 483( )
Page 506 and 507:
4.34 Finite Automata Methods 485Ans
Page 508 and 509:
4.34 Finite Automata Methods 487Thi
Page 510 and 511:
4.34 Finite Automata Methods 489fun
Page 512 and 513:
4.34 Finite Automata Methods 491add
Page 514 and 515:
0123012301012323012301234.34 Finite
Page 516 and 517:
4.35 Iterated Function Systems 495a
Page 518 and 519:
4.35 Iterated Function Systems 497T
Page 520 and 521:
4.35 Iterated Function Systems 499o
Page 522 and 523:
4.35 Iterated Function Systems 501T
Page 524 and 525:
4.35 Iterated Function Systems 503p
Page 526 and 527:
4.35 Iterated Function Systems 505i
Page 528 and 529:
4.35 Iterated Function Systems 507E
Page 530 and 531:
4.35 Iterated Function Systems 509a
Page 532 and 533:
4.36 Cell Encoding4.36 Cell Encodin
Page 534 and 535:
5Wavelet Methods5.1 Fourier Transfo
Page 536 and 537:
5.2 The Frequency Domain 515(a)(b)(
Page 538 and 539:
5.2 The Frequency Domain 517and its
Page 540 and 541:
5.3 The Uncertainty Principle 5191
Page 542 and 543:
5.4 Fourier Image Compression 521fi
Page 544 and 545:
5.4 Fourier Image Compression 523st
Page 546 and 547:
5.5 The CWT and Its Inverse 525The
Page 548 and 549:
5.5 The CWT and Its Inverse 527(1)(
Page 550 and 551:
5.5 The CWT and Its Inverse 529(a)5
Page 552 and 553:
5.6 The Haar Transform 53111φ(t)t1
Page 554 and 555:
5.6 The Haar Transform 533after eac
Page 556 and 557:
5.6 The Haar Transform 535The forme
Page 558 and 559:
5.6 The Haar Transform 537procedure
Page 560 and 561:
5.6 The Haar Transform 539Figure 5.
Page 562 and 563:
5.6 The Haar Transform 541and the t
Page 564 and 565:
5.6 The Haar Transform 543123456780
Page 566 and 567:
5.6 The Haar Transform 545clear; %
Page 568 and 569:
5.6 The Haar Transform 547002020404
Page 570 and 571:
5.7 Filter Banks5.7 Filter Banks 54
Page 572 and 573:
5.7 Filter Banks 551Notice how simi
Page 574 and 575:
5.7 Filter Banks 553since each y(n)
Page 576 and 577:
5.7 Filter Banks 555In matrix notat
Page 578 and 579:
5.7 Filter Banks 557stage 3y 0 (n)y
Page 580 and 581:
5.8 The DWT5.8 The DWT 559Informati
Page 582 and 583:
5.8 The DWT 561equations used to ca
Page 584 and 585:
5.8 The DWT 563P5P9P4P10Figure 5.32
Page 586 and 587:
5.8 The DWT 565function wc=fwt2(dat
Page 588 and 589:
5.8 The DWT 567.099305765374 .42421
Page 590 and 591:
5.8 The DWT 5690.10.050.040.0200-0.
Page 592 and 593:
5.8 The DWT 571Nondifferentiable Fu
Page 594 and 595:
5.10 Various Image Decompositions 5
Page 596 and 597:
Page 598 and 599:
Page 600 and 601:
Page 602 and 603:
5.11 The Lifting Scheme 581d n−1
Page 604 and 605:
5.11 The Lifting Scheme 583Given th
Page 606 and 607:
5.11 The Lifting Scheme 585average
Page 608 and 609:
5.11 The Lifting Scheme 587Figure 5
Page 610 and 611:
5.11 The Lifting Scheme 589It is no
Page 612 and 613:
5.12 The IWT 591for l = 0 (this pro
Page 614 and 615:
5.13 The Laplacian Pyramid 593⋄ E
Page 616 and 617:
5.13 The Laplacian Pyramid 595the s
Page 618 and 619:
5.14 SPIHT 597A lossy version of th
Page 620 and 621:
5.14 SPIHT 599most, so a progressiv
Page 622 and 623:
5.14 SPIHT 601of all the wavelet co
Page 624 and 625:
5.14 SPIHT 6031 32 4LL23LH21432HL2H
Page 626 and 627:
5.14 SPIHT 6051. Initialization: Se
Page 628 and 629:
5.14 SPIHT 607Is (1, 4) significant
Page 630 and 631:
5.15 CREW5.15 CREW 609The CREW meth
Page 632 and 633:
5.16 EZW 611(not coordinates) of th
Page 634 and 635:
5.17 DjVu 6134. The 10 is less than
Page 636 and 637:
5.17 DjVu 615decompressed image acc
Page 638 and 639:
5.18 WSQ, Fingerprint Compression 6
Page 640 and 641:
Page 642 and 643:
Page 644 and 645:
5.19 JPEG 2000 6231999, when the JP
Page 646 and 647:
5.19 JPEG 2000 625step two, the wav
Page 648 and 649:
5.19 JPEG 2000 627Vertical offsetIm
Page 650 and 651:
5.19 JPEG 2000 629mallatspaclpacket
Page 652 and 653:
5.19 JPEG 2000 6312 xcbFour rowsFou
Page 654 and 655:
5.19 JPEG 2000 633CoefficientsBitpl
Page 656 and 657:
5.19 JPEG 2000 635obtain the image
Page 658 and 659:
6Video CompressionSound recording a
Page 660 and 661:
6.1 Analog Video 639achieving an ef
Page 662 and 663:
6.1 Analog Video 641employ PAL colo
Page 664 and 665:
6.2 Composite and Components Video
Page 666 and 667:
6.3 Digital Video6.3 Digital Video
Page 668 and 669:
6.3 Digital Video 647munications Co
Page 670 and 671:
6.4 Video Compression 649160352 700
Page 672 and 673:
6.4 Video Compression 651IP B B B P
Page 674 and 675:
6.4 Video Compression 653Previous f
Page 676 and 677:
6.4 Video Compression 655to look fo
Page 678 and 679:
6.4 Video Compression 65714161713 1
Page 680 and 681:
6.4 Video Compression 659−88−7
Page 682 and 683:
6.5 MPEG6.5 MPEG 661Started in 1988
Page 684 and 685:
6.5 MPEG 663Audio layerAudiodecoder
Page 686 and 687:
6.5 MPEG 665Figure 6.18a,b shows th
Page 688 and 689:
6.5 MPEG 667should be assigned a sh
Page 690 and 691:
6.5 MPEG 669Table 6.23 lists the EO
Page 692 and 693:
6.5 MPEG 6719/1 0000 101s 89/2 0000
Page 694 and 695:
6.5 MPEG 673slices. Each slice, in
Page 696 and 697:
6.5 MPEG 675rates. Its eight nonres
Page 698 and 699:
6.5 MPEG 677Increment macroblock In
Page 700 and 701:
6.5 MPEG 679Once the pattern_code b
Page 702 and 703:
power of 2 given by6.5 MPEG 681f =2
Page 704 and 705:
6.6 MPEG-4 683normally with few non
Page 706 and 707:
6.6 MPEG-4 685sion. Imagine a case
Page 708 and 709:
6.6 MPEG-4 687also includes specifi
Page 710 and 711:
6.7 H.261 689layers, and macroblock
Page 712 and 713:
7Audio CompressionText does not occ
Page 714 and 715:
7.1 Sound 693The problem with measu
Page 716 and 717:
7.2 Digital Audio 695cannon at muzz
Page 718 and 719:
7.2 Digital Audio 697these samples
Page 720 and 721:
7.3 The Human Auditory System 69940
Page 722 and 723:
7.3 The Human Auditory System 701wi
Page 724 and 725:
7.3 The Human Auditory System 703fa
Page 726 and 727:
encoder uses the logarithmic expres
Page 728 and 729:
7.4 µ-Law and A-Law Companding 707
Page 730 and 731:
7.4 µ-Law and A-Law Companding 709
Page 732 and 733:
7.5 ADPCM Audio Compression 711X[n]
Page 734 and 735:
7.6 MLP Audio 713startsample
Page 736 and 737:
7.6 MLP Audio 715The main features
Page 738 and 739:
7.7 Speech Compression 717and its e
Page 740 and 741:
7.7 Speech Compression 719There are
Page 742 and 743:
7.7 Speech Compression 721speech pr
Page 744 and 745:
7.7 Speech Compression 723can be wr
Page 746 and 747:
7.8 Shorten7.8 Shorten 725Shorten i
Page 748 and 749:
7.8 Shorten 727ŝ(t) =3s(t − 1)
Page 750 and 751:
7.9 MPEG-1 Audio Layers7.9 MPEG-1 A
Page 752 and 753:
7.9 MPEG-1 Audio Layers 731Bitrate
Page 754 and 755:
7.9 MPEG-1 Audio Layers 733and prod
Page 756 and 757:
7.9 MPEG-1 Audio Layers 73532 sampl
Page 758 and 759:
7.9 MPEG-1 Audio Layers 737audio da
Page 760 and 761:
7.9 MPEG-1 Audio Layers 739Layer I:
Page 762 and 763:
7.9 MPEG-1 Audio Layers 741110quant
Page 764 and 765:
7.9 MPEG-1 Audio Layers 743The bit
Page 766 and 767:
7.9 MPEG-1 Audio Layers 745Table 7.
Page 768 and 769:
7.9 MPEG-1 Audio Layers 747numberof
Page 770 and 771:
7.9 MPEG-1 Audio Layers 749The MDCT
Page 772 and 773:
7.9 MPEG-1 Audio Layers 75118X 17X
Page 774 and 775:
7.9 MPEG-1 Audio Layers 753Startyes
Page 776 and 777:
8Other MethodsPrevious chapters dis
Page 778 and 779:
8.1 The Burrows-Wheeler Method 7571
Page 780 and 781:
8.1 The Burrows-Wheeler Method 759A
Page 782 and 783:
8.2 Symbol Ranking 761String S cont
Page 784 and 785:
8.2 Symbol Ranking 763each failure
Page 786 and 787:
8.3 ACB 765Figure 8.6. Two linked l
Page 788 and 789:
8.3.1 The Encoder8.3 ACB 767The cur
Page 790 and 791:
8.3 ACB 769It is clear that ACB is
Page 792 and 793:
8.3 ACB 771followed by b =1)thenq =
Page 794 and 795:
8.4 Sort-Based Context Similarity 7
Page 796 and 797:
8.4 Sort-Based Context Similarity 7
Page 798 and 799:
λH0 ↔↓#98.5 Sparse Strings 777
Page 800 and 801:
8.5 Sparse Strings 779The decoder w
Page 802 and 803:
8.5 Sparse Strings 781is in class i
Page 804 and 805:
8.5 Sparse Strings 783The R i codes
Page 806 and 807:
8.5 Sparse Strings 785matrix elemen
Page 808 and 809:
8.5 Sparse Strings 787This is why P
Page 810 and 811:
8.6 Word-Based Text Compression 789
Page 812 and 813:
8.6 Word-Based Text Compression 791
Page 814 and 815:
8.7 Textual Image Compression 793it
Page 816 and 817:
8.7 Textual Image Compression 795re
Page 818 and 819:
encoded with a probability8.7 Textu
Page 820 and 821:
8.8 Dynamic Markov Coding 799◦◦
Page 822 and 823:
8.8 Dynamic Markov Coding 801probab
Page 824 and 825:
8.8 Dynamic Markov Coding 803Assign
Page 826 and 827:
8.8 Dynamic Markov Coding 805001001
Page 828 and 829:
8.8 Dynamic Markov Coding 807(a)Sta
Page 830 and 831:
8.9 FHM Curve Compression 809no poi
Page 832 and 833:
8.10 Sequitur 811of Huffman codes c
Page 834 and 835:
8.10 Sequitur 813A. Figure 8.43c sh
Page 836 and 837:
8.10 Sequitur 815As mentioned earli
Page 838 and 839:
8.11 Triangle Mesh Compression: Edg
Page 840 and 841:
Page 842 and 843:
Page 844 and 845:
Page 846 and 847:
Page 848 and 849:
8.12 SCSU: Unicode Compression 827T
Page 850 and 851:
8.12 SCSU: Unicode Compression 8291
Page 852 and 853:
8.12 SCSU: Unicode Compression 8312
Page 854 and 855:
8.12 SCSU: Unicode Compression 833T
Page 856 and 857:
BibliographyAll URLs have been chec
Page 858 and 859:
Bibliography 837Burt, Peter J., and
Page 860 and 861:
Bibliography 839Dewitte, J., and J.
Page 862 and 863:
Bibliography 841Hafner, Ullrich (19
Page 864 and 865:
Bibliography 843Jarvis, J. F., and
Page 866 and 867:
Bibliography 845Marking, Michael P.
Page 868 and 869:
Bibliography 847Pennebaker, William
Page 870 and 871:
Bibliography 849Sacco, William, et
Page 872 and 873:
Bibliography 851Udupa, Raghavendra
Page 874 and 875:
Bibliography 853Yokoo, Hidetoshi (1
Page 876 and 877:
GlossaryACB. A very efficient text
Page 878 and 879:
Glossary 857Block Decomposition. A
Page 880 and 881:
Glossary 859Conditional Image RLE.
Page 882 and 883:
Glossary 861Dictionary-Based Compre
Page 884 and 885:
Glossary 863EZW. A progressive, emb
Page 886 and 887:
Glossary 865Huffman Coding. A popul
Page 888 and 889:
Glossary 867coding, and uses quanti
Page 890 and 891:
Glossary 869LZSS. This version of L
Page 892 and 893:
Glossary 871Prefix Compression. In
Page 894 and 895:
Glossary 873SCSU. SCSU is a new com
Page 896 and 897:
Glossary 875data by applying the op
Page 898 and 899:
Joining the DataCompression Communi
Page 900 and 901:
IndexThe index caters to those who
Page 902 and 903:
Index 881binary tree predictive cod
Page 904 and 905:
Index 883in MPEG audio, 736, 743, 7
Page 906 and 907:
Index 885ear (human), 698-701Eastma
Page 908 and 909:
Index 887grayscale image, 252, 260,
Page 910 and 911:
Index 889recommendation CD 14495, 3
Page 912 and 913:
Index 891of DMC, 801, 804of JBIG, 3
Page 914 and 915:
Index 893lossy option, 352, 415medi
Page 916 and 917:
Index 895soundfricative, 719plosive
Page 918 and 919:
Index 897Udupa, Raghavendra, xiii,
Page 920:
ColophonThe first edition of this b
show all

Data Compression: The Complete Reference

Create successful ePaper yourself

Delete template?

Save as template?