Image Processing Architecture, © Oleh TretiakPage 1Lecture 10 ECEC 453 Image Processing Architecture Lecture 10, 2/17/2004 MPEG-2, Industrial Strength Video Compression and Friends Oleh Tretiak Drexel University

Image Processing Architecture, © Oleh TretiakPage 2Lecture 10 Lecture Outline Basic Video Coding Features of MPEG-1 Features of H261 MPEG-2 Introduction to MPEG-4

Image Processing Architecture, © Oleh TretiakPage 3Lecture 10 Basic Video Coding Layered digital video stream Picture types Coding parameters and compression Decoder: general diagram Options for block coding

Image Processing Architecture, © Oleh TretiakPage 4Lecture 10 Picture of Layers

Image Processing Architecture, © Oleh TretiakPage 5Lecture 10 Video Compression: Picture Types Group of Pictures: Three types  I — intraframe coding only  P — predictive coding  B — bi-directional coding

Image Processing Architecture, © Oleh TretiakPage 6Lecture 10 Typical MPEG coding parameters Typical sequence  IPBBPBBPBBPBBPBB (16 frames)

Image Processing Architecture, © Oleh TretiakPage 7Lecture 10 Block Diagram of MPEG Decoder I frame P frame B frame

Image Processing Architecture, © Oleh TretiakPage 8Lecture 10 Macroblock Coding: I & P I pictures (almost like JPEG)  Divided into slices and macroblocks  No motion compensation  Each macroblock can have different quantization  DC and AC coded differently, as in JPEG  Different coding tables from JPEG P pictures  Divided into slices and macroblocks  Option: no motion compensation  Option: can code block as inter or intra (like I picture)  Can skip macroblock (replace with previous). Great compression

Image Processing Architecture, © Oleh TretiakPage 9Lecture 10 Coding Image Blocks B pictures  Inter or intra?  Forward, backward, interpolational?  Code block or skip?  Quantization step? Statistics for an image sequence

Image Processing Architecture, © Oleh TretiakPage 10Lecture 10 Old standards: MPEG-1 and Videoconferencing

Image Processing Architecture, © Oleh TretiakPage 11Lecture 10 MPEG-1: ‘1.5’ Mbps Sample rate reduction in spatial and temporal domains Spatial  Block-based DCT  Huffman coding (no arithmetic coding) of motion vectors and quantized DCT coefficients  352 x 340 pixels, 12 bits per pixel, picture rate 30 pictures per second —> 30.4 Mbps  Coded bit stream 1.15 Mbps (must leave bandwidth for audio)  Compression 26:1  Quality better than VHS! Temporal  Block-based motion compensation  Interframe coding (two kinds)

Image Processing Architecture, © Oleh TretiakPage 12Lecture 10 Video Teleconferencing Comprehensive Standard: H.320 Components of H.320  H.261: Video coding, 64 to 1920 kbits/sec  G.722, G.726, G.728: Audio coding from 16 kbits/sec to 64 kbits/sec  H.221: Multiplexing of audio and video (frame based rather than packet based)  H.230 and H.242: Handshaking and control  H.233: encryption

Image Processing Architecture, © Oleh TretiakPage 13Lecture 10 Generic Video Telephone System

Image Processing Architecture, © Oleh TretiakPage 14Lecture 10 H.261 Features Common Interchange Format  Interoperability between 25 fps and 30 fps countries  252 pix/line, 288 line, 30 fps noninterlace  Terminal equipment converts frame and line numbers  Y Cb Cr components, color sub-sampled by a factor of 2 in both directions Coding  DCT, 8x8, 4 Y and 2 chrominance per masterblock  I and P frames only, P blocks can be skipped  Motion compensation optional, only integer compensation  (Optional) forward error correction coding

Image Processing Architecture, © Oleh TretiakPage 15Lecture 10 H.261 vs MPEG-1 Similarities  CIF, SIF, non-interlaced  DCT technology Differences  H.261 uses mostly P frames, no B frames  H.261 typical bit rates much lower (down to 64 kbits/sec)  Low bit rates achieved by reducing frame rate  Simpler motion compensations  End-to-end coding delay must be low Conclusion: Same technology, different design to meet different needs

Image Processing Architecture, © Oleh TretiakPage 16Lecture 10 MPEG 2 i, i = 0, 1 History & Goals Expanding universe of video coding What are MPEG-2 profiles? Features of MPEG-2

Image Processing Architecture, © Oleh TretiakPage 17Lecture 10 MPEG Home Official web site  ( still works) (  Information site  (unchanged) History  MPEG-1, the standard for storage and retrieval of moving pictures and audio on storage media (approved Nov. 92)  MPEG-2, the standard for digital television (approved Nov. 94)  MPEG-4 version 1, the standard for multimedia applications (approved Oct. 98), version 2, (approved Dec. 99)  Under development:  MPEG-4 versions 3&4  MPEG-7 the content representation standard for multimedia information search, filtering, management and processing.  Started MPEG-21, the multimedia framework.

Image Processing Architecture, © Oleh TretiakPage 18Lecture 10 MPEG Example Film on DVD: 8 Gbytes Playing time: 2 hours Bit rate 8e9 bytes x 8 bits/byte / 7200 seconds ~ 9 Mbits/sec Information? on the web  o.asp o.asp  ‘ Bit Rate Explained Bit rate describes how much information there is per second in a stream of data. You might have seen audio files described as “128–Kbps MP3” or “64–Kbps WMA.” Kbps stands for “kilobytes per second,”....’  Site claims that 64 Kbps WMA is as good as 128 Kbps MP3  Ignorance about bits and bytes does not encourage credibility

Image Processing Architecture, © Oleh TretiakPage 19Lecture 10 MPEG-2 Goals Compatibility with MPEG-1 Good picture quality Flexibility in input format Random access capability (I pictures) Capability for fast forward, fast reverse play, stop frame Bit stream scalability Low delay for 2-way communications (videoconferencing) Resilience to bit errors

Image Processing Architecture, © Oleh TretiakPage 20Lecture 10 MPEG-2 Implications No reason to restrict to CCIR 601  High resolution can be included (HDTV) No single standard can satisfy all requirements  Family of standards Most applications use a small set of the features  Toolkit approach

Image Processing Architecture, © Oleh TretiakPage 21Lecture 10 MPEG-2 profiles A profile is a subset of the entire MPEG-2 bit-stream syntax  Simple  Main  4:2:2  SNR  Spatial  High  Multiview Each profile has several levels (resolution quality)  Low — MPEG1  Main — CCIR 601  High-1440 (Video Editing)  High (HDTV)

Image Processing Architecture, © Oleh TretiakPage 22Lecture 10 Features of MPEG-2 Support of both non-interlaced and interlaced pictures Color handling  Y Cb Cr color space  Several subsampling schemes are used  4:2:0, 4:2:2, 4:4:4 MPEG-2 sequence can be either frames or fields  Both frame prediction and field prediction are supported  There can be motion between two fields in a frame, so that frame prediction is more tricky  In frame prediction, both fields constitute one picture  In field prediction, either field in the previous frame or the previous field in this frame can be used as reference  Robustified coding of motion vectors to protect against bit errors  Special prediction modes: 16x8, dual-prime

Image Processing Architecture, © Oleh TretiakPage 23Lecture 10 MPEG-2: DCT and Quantization Two quantizers: one for intra blocks and one for non-intra blocks Support different quantization blocks for luminance and chrominance Scalable bit streams  data partitioning, SNR scalability, temporal scalability, spatial scalability  Data partitioning: headers and motion vectors in two bit streams  SNR scalability: lower layer provided basic video, other layers provide enhancements. Basic layer sent with robust modulation  Spatial scalability: lower layer provides basic resolution (e. g., MPEG-1), upper layer provides detail  Temporal scalability: lower layer provides basic (low) frame rate

Image Processing Architecture, © Oleh TretiakPage 24Lecture 10 MPEG-2: Profiles 4:2:2 profile at Main level  Two Y blocks for each pair of Cb, Cr blocks  Distribution format for video production  Robust for several compressions and decompressions  720x608, 30 fps  50 Mbit/sec  Luminance full raster, chrominance are at full line rate  DC precision of intra blocks can be up to 11 bits Main (4:2:0) profile at Main level  Four Y blocks for each pair of Cb, Cr blocks  Intended for broadcast quality (actually, is better)  15 Mbit/sec Main profile at low level  Like MPEG-1

Image Processing Architecture, © Oleh TretiakPage 25Lecture 10 MPEG2 features Schemes for ‘frame’ and field coding. There are two fields in a frame, T (top) B (bottom) Either can be first  Frame prediction for frame pictures  What’s there to say?  Field prediction for field pictures  Target macroblock is in one field  Prediction pixels come from one field  Can be the same of different parity as target field  Field prediction for frame pictures  Dual prime for P-pictures  16x8 macroblock for field pictures Motion vectors coded at half-pel resolution

Image Processing Architecture, © Oleh TretiakPage 26Lecture 10 MPEG2 - Alternate Scan Zig-zag scanAlternate scan

Image Processing Architecture, © Oleh TretiakPage 27Lecture 10 MPEG2 — Subsampling Suppose picture is 720x480  4:4:4  Luminance and 720x480  4:2:2  720x480, chrominance 360x480  4:2:0  Luminance 420x480, chrominance 360x240 Weird terminology

Image Processing Architecture, © Oleh TretiakPage 28Lecture 10 Low Y ~ 352x240 Cb, Cr ~ 176x pictures per second +/- 64 pixel displacement, half pixel resolution

Image Processing Architecture, © Oleh TretiakPage 29Lecture 10 Main (4:2:0) Y ~ 720x480 Cb,Cr ~ 360x frames per second 4:3, 16:9 aspect ratio Bitrate 15 Mbps (some applications as low as 5 Mbps) Digital television

Image Processing Architecture, © Oleh TretiakPage 30Lecture 10 High Y 1920x1152 Cb, Cr 960x frames per second 80 Mbps HDTV

Image Processing Architecture, © Oleh TretiakPage 31Lecture 10 Low rate Where is it needed? How is it done?

Image Processing Architecture, © Oleh TretiakPage 32Lecture 10 MPEG-2: DCT and Quantization Two quantizers: one for intra blocks and one for non-intra blocks Support different quantization blocks for luminance and chrominance Scalable bit streams  data partitioning, SNR scalability, temporal scalability, spatial scalability  Data partitioning: headers and motion vectors in two bit streams  SNR scalability: lower layer provided basic video, other layers provide enhancements. Basic layer sent with robust modulation  Spatial scalability: lower layer provides basic resolution (e. g., MPEG-1), upper layer provides detail  Temporal scalability: lower layer provides basic (low) frame rate

Image Processing Architecture, © Oleh TretiakPage 33Lecture 10 MPEG-4 Multimedia Standard Thumbnail Description

Image Processing Architecture, © Oleh TretiakPage 34Lecture 10 What Is Left for MPEG-4? Initial goals  Coding standards for lower-than-MPEG-1 rates  Hidden agenda: Incorporate new coding methods  Wavelet, fractal  Revised agenda: Object-based coding MPEG-4 Architecture  Input to coder consist of audio, video, and stored objects  Decoder combines encoded objects with local objects  Example: send text by sending character codes, receiver uses character generator.

Image Processing Architecture, © Oleh TretiakPage 35Lecture 10 Schematic Overview of MPEG-4

Image Processing Architecture, © Oleh TretiakPage 36Lecture 10 MPEG-4 Ideas Video Object Plane (VOP)  A VOP can be a natural image from video camera or from a graphics database  A VOP can consist of several visual object. Visual objects do not have to have rectangular outline (arbitrary shape)  A scene consists of several VO’s and VOP’s with appropriate compositing  Different VOP’s can have their own motion In principle, a visual scene can be decomposed into video objects by segmentation. Color and texture can be attributes of visual objects A viewer can manipulate VO’s.