Image Compression

by

Mohamed Hagras
Contents

 What is Image Compression?

 Why Image Compression?
 Lossless and Lossy Techniques
 Bitmap Format
 Graphics Interchange Format
 Portable Network Graphics Format
 Tag Image File format
 JPEG
 JPEG 2000
22/05/2010 Image Compression 2
Contents

 Video Compression Principals

 Frames
 Types of frames
• I-frames
• P-frames
• B-frames
• PB-frames
• D-frames
 Conclusion

22/05/2010 Image Compression 3

What is Image Compression?
 A technique used to reduce the volume of information
to be transmitted about an image.
Width
1 Unit

M=Number
Height 1 of Samples
Unit M D=Spacing
D
Between
Samples

1/2
M
22/05/2010 Image Compression 4
Why Image Compression?
 Requirements may outstrip the anticipated increase of
storage space and bandwidth.

 For data storage and data transmission:

 DVD
 Real time applications
 Printer

 Take advantage of :
 Spatial redundancy
 Temporal redundancy (videos)
22/05/2010 Image Compression 5
Lossless And Lossy Compression
Lossless Compression Lossy Compression
Definition  There is no information loss.  Information loss is
 The image can be tolerable.
reconstructed exactly the same
as the original.
e.g RLE JPEG 2000
Applications  Medical imagery  Commercial distribution
 Archiving (DVD).
 Rate constrained
environment where
lossless methods can not
provide enough
compression ratio.

22/05/2010 Image Compression 6

Bitmap Format
 Files with .bmp extension
 Each pixel is represented by a fixed number of bits
(integral power of 2)
Advantages Disadvantages
 Simple  Minimum or zero
 Widely used compression
 Relatively well
documented
 Free of patents

22/05/2010 Image Compression 7

Graphics Interchange Format

 GIF supports transparency and interlacing

 Transparency is supported by specifying which pixels of the image is
unused and as such, to reuse the background color.
 Interlacing creates the illusion of faster loading graphics as the
image is presented to the viewer in a series of interlaced frames.

 256 colors out of a palette of 224 colors are indexed.

 Amongst the oldest formats available.
 Supports animations.

22/05/2010 Image Compression 8

Portable Network Graphics Format
 PNG Developed to improve upon and replace GIF.
 Supports:
 2D Interlacing
 Transparency
 Variable Color Depth
 Uses Lossless Data compression method called DEFLATE.

The same
A PNG image
image is
with an 8-bit
overlaid
transparency
onto a
channel
checkered
background

 The only Disadvantage of png is that it doesn’t support Animations.

22/05/2010 Image Compression 9
Tag Image File Format
 Developed by Aldus and Microsoft.
 Tags used to indicate how data is arrange and
compressed.
 e.g LZW/RLE compression
Advantages Disadvantages
 Highly flexible and  TIFF image can’t have
platform independent more than 4GB of data
 Widely used by  Difficult to write TIFF
scanning,faxing, word readers which take
processing advantage of all its features
 capable of describing
bilevel, grayscale, palette-
color, and full-color image
data

22/05/2010 Image Compression 10

JPEG
 Joint Photographic Expert’s Group.
 International standard for photographs.
 Lossless/lossy.
 Based on the facts that :
 Humans are more sensitive to lower spatial frequency
components.
 A large majority of useful image contents change
relatively slowly across images.

22/05/2010 Image Compression 11

JPEG Encoding
Cr
Cb f(i, j) F(u, v) Fq(u, v)
Y DCT Quantization
8x8 8x8

Quant…
Tables
Coding Zig Zag
Tables Scan

Header
Tables
DPCM
Data Entropy
Coding
RLC

22/05/2010 Image Compression 12

JPEG (Cont)
Advantages Disadvantages
Compression ratios of 20:1 Doesn’t support
are easily attained. transparency.

24-bits per pixel can be Doesn’t work well with

used leading to better sharp edges.
accuracy.

Progressive Almost always lossy and

JPEG(interlacing) No target bit rate

22/05/2010 Image Compression 13

Comparative Analysis

a. 256 color bmp b. gif c. jpeg

253 KB 4.08 KB 10.1 KB

d. Png e. tiff
6.27 K 16.2 KB
22/05/2010 Image Compression 14
JPEG 2000
 Wavelet based image compression standard

Source
Image Data
Forward Entropy
Quantization Compressed Image Data
Transform Encoding

Reconstructed
Image
Data
Inverse Inverse Entropy
Compressed Image Data
Transform Quantization Decoding

22/05/2010 Image Compression 15

Wavelet Transform Of An Image

Advantages:

 Lossless and lossy

compression.
 Progressive transmission by
pixel accuracy and
resolution.
 Region-of-Interest Coding.
 Random codestream access
and processing.
 Robustness to bit-errors.
 Content-based description.
 Side channel spatial
information (transparency).

22/05/2010 Image Compression 16

JPEG vs. JPEG 2000

512x512 image reconstructed after compression of 0.2 bpp

using JPEG and JPEG 2000 respectively

22/05/2010 Image Compression 17

Moving Images (Videos)
 Any video can be thought up of as a sequence of
digitized pictures or frames.
Types Of Frame Coding
Intra-Frame Coding Inter-Frame Coding
(Spatial Redundancy) (Temporal Redundancy)

 I-Frame  P-Frame
 B-Frame

 JPEG algo applied to each frame  Only a small portion of each

independently moving JPEG or frame is involved with any
MJPEG motion, so only info related to
 Typical compression ratios bw those segments is send
10:1 and 20:1
 Not large enough to produce
the needed compression ratios

22/05/2010 Image Compression 18

 I-Frame P-Frame B-Frame PB-Frame
 Each frame is  Encoded using a  Their contents  Two neighboring
coded combination of are predicted P- and B-frames
independently. either a preceding using both past are encoded as if
 Y (luminance) I-frame or P- and future frames they were a single
and Cb, Cr frame.  Allows for frame.
(chrominance)  No. of P-frames occasional fast  It increases the
matrices encoded limited as errors moving objects frame rate
separately using propagate from  Provides better without
JPEG algo. one p-frame to motion increasing the
 I-frames another. estimation. resulting bit rate
frequently used  M = prediction required.
 Provides highest
in order to tackle span = no. of level of
corruption. frames bw a P- compression.
 N = GOP = no. of frame and the
immediately  As they are not
frames bw involved in
successive I- preceeding I-
frame or P-frame. coding of other
frames, frames they do
range(3,12). not propagate
errors.

22/05/2010 Image Compression 19

Encoding Operation

22/05/2010 Image Compression 20

D-Frame
 Used in movie / video-on-demand applications.

 User may wish to rewind or fast-forward the video.

 This requires the compressed video to be

decompressed at much higher speeds.

 D-frames inserted at regular intervals throughout the

stream.

 By using only the encoded DC coefficients of each

block of pixels in the periodically inserted D-frames, a
low resolution sequence of frames is provided at much
higher speeds.
22/05/2010 Image Compression 21
Performance
 Compression ratios for I-frames is similar to that
obtained with JPEG and is typically between 10:1 and
20:1 depending on the complexity of the frame
contents.

 Compression ratios for P-frames and B-frames are

higher and depend on the search algorithm used.
Typical figures are 30:1 for P-frames and 30:1 through
50:1 for B-frames.

22/05/2010 Image Compression 22

Conclusion
 Every format has its own advantages and disadvantages.
 Depending upon the application the appropriate format
should be used.
 Bmp is simple but heavy.
 Only redeeming feature of gif : supports animations.
 Png better than the rest in overall terms.
 Tiff is appropriate where flexibility is needed.
 JPEG is good for photography.
 JPEG 2000 is an all encompassing standard.

22/05/2010 Image Compression 23

Thanks For ListeninG