13 Image Compression

Embed Size (px)

Citation preview

  • 7/31/2019 13 Image Compression

    1/21

    7/14/2012 Image Compression 1

    Image Compression

  • 7/31/2019 13 Image Compression

    2/21

    7/14/2012 Image Compression 2

    Reference

    [1] Gonzalez and Woods, Digital Image

    Processing.

  • 7/31/2019 13 Image Compression

    3/21

    7/14/2012 Image Compression 3

    Objective

    Reduce the number of bytes required to

    represent a digital image

    Redundant data reduction

    Remove patterns

    Uncorrelated data confirms redundant data

    elimination Auto correlation?

  • 7/31/2019 13 Image Compression

    4/21

    7/14/2012 Image Compression 4

    Enabling Technology

    Compressions is used in

    FAX

    RPV

    Teleconference

    REMOTE DEMO

    etc

  • 7/31/2019 13 Image Compression

    5/21

    7/14/2012 Image Compression 5

    Review

    What and how to exploit data redundancy

    Model based approach to compression

    Information theory principles

    Types of compression

    Lossless, lossy

  • 7/31/2019 13 Image Compression

    6/21

    7/14/2012 Image Compression 6

    Information recovery

    We want to recover the information, with reduceddata volumes.

    Reduce data redundancy.

    How to measure the data redundancy.

    ProcessingData Information

  • 7/31/2019 13 Image Compression

    7/21

    7/14/2012 Image Compression 7

    Relative Data Redundancy

    Assume that we have two data sets D1 and D2.

    Both on processing yield the same information.

    Let n1 and n2 be the infocarrying units of therespective data sets.

    Relative data redundancy is defined on comparing the

    relative dataset sizes

    RD = 11/CR

    where CR is the compression ratio

    CR = n1/ n2

  • 7/31/2019 13 Image Compression

    8/21

    7/14/2012 Image Compression 8

    Examples

    RD = 11/CR

    CR = n1/ n2

    D1 is the original and D2 is compressed.

    When CR = 1, i.e. n1 = n2 then RD=0; no data

    redundancy relative to D1 .

    When CR = 10, i.e. n1 = 10 n2 then RD=0.9; impliesthat 90% of the data in D1 is redundant.

    What does it mean if n1

  • 7/31/2019 13 Image Compression

    9/21

  • 7/31/2019 13 Image Compression

    10/21

    7/14/2012 Image Compression 10

    Coding Redundancy

    How to assign codes to alphabet

    In digital image processing

    Code = gray level value or color value Alphabet is used conceptually

    General approach

    Find the more frequently used alphabet

    Use fewer bits to represent the more frequently used

    alphabet, and use more bits for the less frequently used

    alphabet

  • 7/31/2019 13 Image Compression

    11/21

    7/14/2012 Image Compression 11

    Coding Redundancy 2

    Focus on gray value images

    Histogram shows the frequency of occurrence of aparticular gray level

    Normalize the histogram and convert to a pdfrepresentationlet rkbe the random variable

    pr(rk) = nk/n ; k = 0, 1,2 ., L-1, where L is the number of gray levelvalues

    l(rk) = number of bits to represent rk

    Lavg = k=0 to L-1 l(rk) pr(rk) = average number of bits to encode onepixel. For M x N image, bits required is MN Lavg

    For an image using an 8 bit code, l(rk) = 8, Lavg = 8.

    Fixed length codes.

  • 7/31/2019 13 Image Compression

    12/21

    7/14/2012 Image Compression 12

    Fixed vs Variable Length Codes

    From [1]

    Lavg = 2.7

    CR= 3/2.7 = 1.11

    RD = 11/1.11 = 0.099

  • 7/31/2019 13 Image Compression

    13/21

    7/14/2012 Image Compression 13

    Code assignment view

    From [1]

  • 7/31/2019 13 Image Compression

    14/21

    7/14/2012 Image Compression 14

    Interpixel Redundancy

    From [1]

  • 7/31/2019 13 Image Compression

    15/21

    7/14/2012 Image Compression 15

    Run Length Coding

    From [1]

    CR=1024*343/12166*11

    = 2.63

    RD = 1-1/2.63 = 0.62

  • 7/31/2019 13 Image Compression

    16/21

    7/14/2012 Image Compression 16

    Psychovisual Redundancy

    Some visual characteristics are less

    important than others.

    In general observers seeks out certain

    characteristicsedges, textures, etcand

    the mentally combine them to recognize the

    scene.

  • 7/31/2019 13 Image Compression

    17/21

    7/14/2012 Image Compression 17

    From [1]

  • 7/31/2019 13 Image Compression

    18/21

    7/14/2012 Image Compression 18

    From [1]

  • 7/31/2019 13 Image Compression

    19/21

    7/14/2012 Image Compression 19

    Fidelity Criteria

    Subjective

    Objective

    Sum of the absolute error

    RMS value of the error

    Signal to Noise Ratio

  • 7/31/2019 13 Image Compression

    20/21

    7/14/2012 Image Compression 20

    Subjective scale

    From [1]

  • 7/31/2019 13 Image Compression

    21/21

    7/14/2012 Image Compression 21

    Image Compression Model

    Run length JPEG Huffman

    From [1]