Special Module on Media Processing

and Communication

Dayalbagh Educational Institute Indian Institute of Technology Delhi

(DEI) (IITD)
Dayalbagh Agra New Delhi
 Recap

Lecture 1

Overview

Digital Representation
• Audio
• Image
• Video
• Geometry

Need of Compression

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 2

 Image Compression
Compression Ratio
Cr = no/nc

no = Number of carrying units (bits) in the original

data (image)
nr = Number of carrying units (bits) in the compressed
data (image)

Also,
Rd = 1 – 1/ Cr
Rd = Relative data redundancy

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 3

 Image Compression
Fidelity Criteria

Measure of loss or degradation

• Mean Square Error (MSE)

• Signal to Noise Ratio (SNR)

• Subjective Voting

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 4

 Compression
Compression Techniques

• Loss-less Compression
Information can be compressed and
restored without any loss of information
• Lossy Compression
Large compression, perfect recovery is
not possible

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 5

 Compression
Compression Techniques

Symmetric
• Same time for compression (coding) and
decompression (decoding)
• Used for dialog (interactive) mode applications
Asymmetric
• Compression is done once so can take longer
• Decompression is done frequently so should be fast
• Used for retrieval model applications

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 6

 Image Compression
Data Redundancy

• Coding
Variable length coding with shorter codes
for frequent symbols
• Interpixel
Neighboring pixels are similar

• Psychovisual
Human visual perception - limited

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 7

 Image Compression
Coding Redundancy
Example: (from Digital Image Processing by Gonzalez and Woods)

fixed length coding variable length coding

Avg length=3 bits Avg length=2.7 bits

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 8

 Image Compression
Interpixel Redundancy
Example: (from Digital Image Processing by Gonzalez and Woods)

Image

Histogram

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 9

 Image Compression
Interpixel Redundancy
Example: (from Digital Image Processing by Gonzalez and Woods)

Image
High interpixel correlation

Histogram

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 10

 Image Compression
Psychovisual Redundancy
Example: (from Digital Image Processing by Gonzalez and Woods)

Original 256 levels 16 level quantization IGS quantization

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 11
 Image Compression
Loss-less Techniques

• Coding redundancy
Variable length coding

• Interpixel redundancy
Run length coding
Predictive coding

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 12

 Image Compression
Variable Length Coding (Huffman Coding)

Sequence of symbols (a1, a2, a3, a4, a5) with associated

probabilities (p1, p2, p3, p4, p5)

• Start with two symbols of the least probability

a1:p1
a2:p2
• Combine (a1 or a2) with probability (p1+p2)
• Do it recursively (sort and combine)
• A binary tree construction

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 13

 Image Compression
Variable Length Coding (Huffman Coding)
Example:
Symbols and their probabilities of occurrence
a1 (0.2), a2 (0.4), a3 (0.2), a4 (0.1), a5 (0.1)
a2 (0.4)

a1(0.2)
Sort in
probability a3(0.2)

a4(0.1)

a5(0.1)
Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 14
 Image Compression
Variable Length Coding (Huffman Coding)
Example:
Sort

a2 (0.4)

a1(0.2)

a3(0.2)

a4(0.1)

a5(0.1)

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 15

 Image Compression
Variable Length Coding (Huffman Coding)
Example:
Sort combine

a2 (0.4)

a1(0.2)

a3(0.2)

a4(0.1) 0.2

a5(0.1)

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 16

 Image Compression
Variable Length Coding (Huffman Coding)
Example:
Sort combine Sort

0.4
a2 (0.4)
0.2
a1(0.2)
0.2
a3(0.2)

a4(0.1) 0.2 0.2

a5(0.1)

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 17

 Image Compression
Variable Length Coding (Huffman Coding)
Example:
Sort combine Sort combine Sort combine Sort combine

0.4 0.4 0.6

a2 (0.4) 1
0.2 0.4
a1(0.2) 0.6 0.4
0.2 0.2
a3(0.2) 0.4

a4(0.1) 0.2 0.2

a5(0.1)

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 18

 Image Compression
Variable Length Coding (Huffman Coding)
Example:
Sort combine Sort combine Sort combine Sort combine

1 0.4 1 0.4 1 0.6 0

1 a2 (0.4) 1
01 0.2 01 0.4 00
01 a1(0.2) 0.6 0.4 1
000 0.2 000 0.2
000 a3(0.2) 0.4
0010 01
0010 a4(0.1) 0.2 0.2
001
0011 a5(0.1)
0011 Assign code

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 19

 Image Compression
Decoding
Example:

00111010001

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 20

 Image Compression
Run Length Coding

Run: a string of the same symbol

Example
input: AAABBCCCCCCCCCAA
output: A3B2C9A2

compression ratio = 16/8 = 2

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 21

 Image Compression
Predictive Coding

Basic premise: Current pixel is similar to the previous pixel

(coherence)
Differential Coding
d(x,y) = I(x,y) – I(x-1,y)

d(x,y) prediction error which is to be encoded.

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 22

 Image Compression
Predictive Coding
Compression

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 23

 Image Compression
Predictive Coding
Decompression

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 24

 Image Compression
Predictive Coding

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 25

 Image Compression
Lossy

• Psychovisual redundancy
• Keep more important information
• Trade off between loss (degradation) and compression

Compression Ratio: 7.7 Compression Ratio: 12.3 Compression Ratio: 33.9

Original

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 26

 Image Compression
Lossy

• Recall Quantization
Discrete value to represent range of values
Irreversible operation
Information loss !

• Predictive Coding
• Transform Coding

Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 27

 Audio
Digital Representation
Audio (Sound): continuous signal (wave form) in time
1D function f(x)

Continuous

Discrete
Slide 7 Lecture 1
Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 28
 Image

2D function f(x,y)

x
Sampling: Discretization in x and y Quantization
Slide 16 Lecture 1
Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 29
 Video
Video is a sequence of images in time

Image
(Frame)

Time

Slide 23 Lecture 1
Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 30
 Graphics

Geometry Data: Meshes

Points

Connectivity

Slide 26 Lecture 1
Special Module on Media Processing and Communication http://www.it.iitd.ac.in/sil864.html Slide 31