1. A seminar on MPEG 1 and 2
Presented By
Anurag Jagetiya
Shiv Prasad Sharma
ME (CSE) Modular-12
Under the guidance of
Ms. Shano Solanki
Astt. Prof (CSE), NITTTR, CHD
MPEG video Compression 1

2.  Need of Video Compression
 Basics of video compression
 MPEG-1
 MPEG-2
 References
Outline
MPEG video Compression 2

3.  Example: HDTV broadcast has resolution of 1920*
1080 at 30 fps using 8 bits to represent primary
colors.
 This leads to a total of 1.5 Gbps data rate
 But, Channel B.W. is only 6MHz that supports around 19.2
Mbps data rate only.
 But, channel also has to support audio, and other data.
 Effective data rate reduced to only 18 mbps
Need of Video Compression
MPEG video Compression 3

4.  Therefore, Compression required is
Compression = 1.5 Gbps / 18mbps = 83
 Hence, Compression ratio required is 83: 1
Contd…
MPEG video Compression 4

5.  Video is made of multiple images called frames.
 Images are having highly redundant data i.e. adjacent
pixels are highly correlated.
 JPEG compresses images.
 Then can’t it be used to compress frames
individually???
Basics of Video Compression
MPEG video Compression 5

6.  ANSWER is YES, we can use jpeg but we can achieve
more compression than that in videos.
MPEG video Compression 6

7. Characteristic of Video
Adjacent frames are similar and changes are due to object
motion: Temporal Correlation
MPEG video Compression 7

8. Key idea
 Predict a new frame from previous frame and specify the
prediction error.
 Prediction error can be coded using image coding methods. (e.g.
JPEG)
 Predictions from past frame are known as forward prediction.
 Prediction error can be coded with fewer bits
 Regions those can’t be predicted well, are coded directly.
MPEG video Compression 8

9.  Previous of next frame is known as reference image.
 Video compression makes use of MOTION
COMPENSATION to predict a frame from the previous
and/ or next frame.
 Motion vectors that describes the transformation from
one image to another that are applied to the target and
synthesized to produce next image.
Contd…
MPEG video Compression 9

10.  Macro Blocks: Compression methods works on a
block of 16*16 pixels called macro blocks.
 I (intra) frames
 independent frames
 Coded without reference to other frames
Key terms
MPEG video Compression 10

11.  P (predictive) Frame
 Not Independent
 These frames are predicted from a past frame (I or P)
 Coded by a forward predictive coding
 Current macro block is predicted from similar macro block in
previous I or P frame. And the difference between macro
blocks is coded
Contd…
MPEG video Compression 11

12. Contd…
MPEG video Compression 12

13.  MPEG introduced a new B frame
 B frame (bi-directional predictive-coded)
 Due to unexpected movement in real scenes, the
target macro block may not have a good matching
in previous frame.
 Therefore, B frame is coded with reference to both
previous and future reference frames (either I or P)
Contd…
MPEG video Compression 13

14. B Frame (Contd…)
MPEG video Compression 14

15.  MPEG-1
 Finalized in 1991
 Initial audio/video compression standard
 Used by VCD’s
 MP3 = MPEG-1 audio layer 3
 Target of 1.5 Mb/s bitrate at 352x240 resolution
 Only supports progressive pictures
MPEG 1: Evolution
MPEG video Compression 15

16. Motion Compensation
MPEG video Compression 16

17.  Chroma (C): Color Information
 Luma (Y): Brightness [Black and white]
 FACT: Human eyes are more sensitive to brightness
information then color.
 Optimization can be achieved by storing more luma
information than chroma.
 Therefore MPEG uses YCbCr model instead of RGB
Chroma Subsampling
MPEG video Compression 17

18.  Y represents Luma
 Cb and Cr jointly represents Chroma
 Cb: Blue color difference
 Cr: Red color difference
 A macro block can be represented in several different
YCbCr color model
 E.g. 4:4:4 means Y=4, Cb=4 and Cr=4 [Full Bandwidth]
 E.g. 4:2:2 means Y=4, Cb=2, Cr=2 [half BW then first one]
YCbCr
MPEG video Compression 18

19. Frame Sequence
Actual frame sequence is determined by encoder and is specified in video
frame header.
Display order: I B B P B B P B B I
Decoding Order: I P B B P B B I B B
NOTE: Decoder can’t work with B frame without its succeeding P or I
frameMPEG video Compression 19

20. Example
MPEG video Compression 20

21.  Different Source format
 CIF: 352x288, SIF: 352x240
 Using bi directional temporal prediction
 More precise motion compensation based predictions.
 Difference in quantization tables
 MPEG allows random access
MPEG 1 Vs H.261
MPEG video Compression 21

22.  In real time transmission of mpeg need of buffering
hence the inevitable delay is an important issue.
 MPEG-1 Can be used to store and play video on CD on
a single computer
 Low bit rate
Issue with MPEG 1
MPEG video Compression 22

23.  Project started in 1990 and finalized in 1994.
 Provided high quality video with bit rate around 4 Mbps
 It was developed as a standard for digital tv broadcast.
 It gained wide popularity in terrestrial, satellite, or cable
network.
 Adopted in DVDs.
MPEG -2
MPEG video Compression 23

24.  Part 1: Combine video and audio data into single/multiple
streams
 Part 2: Offers more advanced video compression tools
 Part 3: Is a multi-channel extension of the MPEG-1 Audio
standard
 Part 4: Specifies protocols of managing MPEG-1 & MPEG-2
bit streams
Parts of MPEG-2
MPEG video Compression 24

25.  Part 5: Specifies a multi-channel audio coding
algorithm
 Part 6 specifies the Real-time Interface (RTI) to
Transport Stream decoders
MPEG video Compression 25

26. MPEG-2 Video Compression
VIDEO STREAM DATA HIRERARCHY
MPEG video Compression 26

27. Four Video Compression Techniques:
1. Pre-processing
2. Temporal Prediction
3. Motion Compensation
4. Quantization
Contd…
MPEG video Compression 27

28. Pre-processing
 Filters out unnecessary information
 Information that is difficult to encode
 Not an important component of human visual perception
MPEG video Compression 28

29. Temporal Prediction:
 Uses the mathematical algorithm Discrete Cosine
Transform (DCT) to:
 Divide each frame into 8X8 blocks of pixels
 Reorganize residual differences between frames
 Encode each block separately
MPEG video Compression 29

30. MPEG video Compression 30

31. MPEG video Compression 31

32. Quantization:
 Lossy Compression
 Compressed range of value by a single quantum value
 Quantization Matrix is designed to provide more
resolution to perceivable frequency
 Set much of the unnecessary elements to Zero
 Uses zig zag ordering and run length encoding.
MPEG video Compression 32

33.  MPEG1 only handles progressive sequences (SIF) MPEG2 is targeted
primarily at interlaced sequences and at higher resolution
 More sophisticated motion estimation methods frame/field prediction
Mode are developed to improve estimation accuracy for interlaced
sequences.
 Different DCT modes and scanning methods are developed for interlaced
sequences.
 MPEG2 has various scalability modes.
 MPEG2 has various profiles and levels, each combination targeted for a
different application
MPEG-2 Vs MPEG-1
MPEG video Compression 33

34.  Submergence
 Handle specific requirements from rapidly developing
multimedia applications
 Advantages over MPEG-1 and MPEG-2
 Object-oriented coding
MPEG 4
MPEG video Compression 34

35.  Another ISO/IEC standard being developed by MPEG
 Content representation standard for information search
 Makes searching the Web for multimedia content as easy as
searching for text-only files
 Operates in both real-time and non real-time
environments
MPEG 7
MPEG video Compression 35

36.  Y. Wang, J. Ostermann, Y. Q. Zhang, Video Processing
and Communications, Prentice Hall, 2002. Chapters
9,11,13
 Fundamental of Multimedia, Li-Drew, Prentice Hall
 John Wiseman, An Introduction to MPEG video
compression
References
MPEG video Compression 36