1 TTM4142 Networked Multimedia Systems Video Basics Image and Video Lossless Compression Leif Arne Rønningen Autumn 2008.

Post on 01-Jan-2016

220 Views

Category:

Documents

2 Downloads

Preview:

Click to see full reader

Transcript

1

TTM4142 Networked Multimedia Systems

Video BasicsImage and Video

Lossless Compression

Leif Arne Rønningen

Autumn 2008

2

Graphics and Image Data Representations

• Large number of (file) formats, PDF, JPEG, GIF, PNG, BMP

• 8 bit Gray-Level Images

• 24 bit Color Images, RGB

• Color Lookup Table - LUT

• Dithering

3

Color in Image and Video

Spectral power distribution of daylight

Wavelength (nm)

4

Human Vision

5

CIE tristimulus values

6

7

Color Models in VideoYUV – PAL and CCIR 601 digital videoYCrCb – JPEG, MPEG, ITU-R 601-4

8

Fundamental Video Concepts

• Component Video, RGB, YUV, YCrCb

• Composite Video, chrominance and luminance signals on single carrier

• S-Video, separated video, one wire for luminance, one wire for composite chrominance

• Analog Video, PAL, interlaced scan, blanking intervals h, v

• Digital Video, YCrCb - chroma subsampling

9

Digital Video

10

HDTV

H.264, (MPEG-4 part 10)

11

Lossless Compression AlgorithmsInformation theoryAlphabet S =

Entropy

12

Run-length Coding

The string AABACCCCCCCCCCDEAB can be coded as

!02ABA!10CDEAB

Which is 14 instead of 18 characters

! Must be reserved

13

14

15

LZW Compression

• With this algorithm, the encoder and decoder build up the same dictionary of recognized strings from a stream dynamically. Each string is given a codeword. The encoder sends short codewords instead of long strings.

• See Li and Drew for details

16

Arithmetic Coding, Sequence to be coded: CAEE$

17

Arithmetic Coding

18

Arithmetic Coding, Decoding

• The coded sequence CAEE$ ends up with the interval [0.33184, 0.3322)

• Find the shortest binary number that is within this interval, 0.01010101 <=> 0.33203125 (procedure 7.2 in book)

• Note that only 8 bits is needed!

• For decoding, go the opposite way of that in the previous slide

• 0.33203125 is in Low-High interval for C

• (0.33203125-0.3)/0.2 = 0.16015624, an A

• and so on

19

Comparing some lossless compression schemes

top related