Multimedia for our time (For Dummies) ISO/IEC Visa Hyoungjune Yi
Preface Environment – TV System – Wireless IS-95, CDMA200, IMT2000 Compression Standard
TV System PAL – Phase Alternation Line rate (Main Europe, inc. Finland) NTSC – National Television Systems Committee (USA & Canada) SECAM – Sequential Coleur Avec Memoire (France, former Soviet) NTSC / PAL / SECAM /
IS-95A, IS-95B, IS-95C, IMT 2000 Why not TDMA? Bandwidth matters 9.6K, 14.4K(IS-95A), 56K, 64K (IS-95B) bps IS-95A,B – Circuit switching CDMA2000 1X (IS-95C), W-CDMA – 144 ~ 384Kbps IMT2000 (International Mobile Telecommunication 2000) – 2 ~ 5 Mbps Bandwidth
Compression Lossy : Reconstructable Lossless Symbol level – Huffman Block level – RLE, Arithmetic Encoding, LZ
Compression Idea – Reduce spacial / temporal redundancies Pre-knowledge – FT, DCT, Haar’s, Gibbs phenomenon, Max-Lloyd Quantizer, Vector quantization – Subband coding and why?, LPF, HPF
Standard Why? – Compatibility – Real player, WMP, Quick-time – Production, cost reduction Standard kills research, right? – Profile level – Decoder only (MPEG2, MPEG4) – Very flexible to encoder side
ISO/IEC IS Lossless 8 X 8 block Mean normalization DCT on each block Quantization (4 quantization matrices per img) Entropy coding of DC component (Zig-zag) Entropy coding of AC component
H.261, H.263 and MPEG H.261 – CCITT, ISDN H.263 – Visual telephony ISO/IEC – CD, 1.5 Mbit/sec ISO/IEC – DVD, over 10Mbit/sec MPEG3 – 50 ~ 100 Mbit/sec ISO/IEC JTC/SC29/WG11 N2323 – Variable bitrate, 9.6K ~ 100Mbps ISO/MPEG N2728, 2729 – Describing multimedia material MPEG11
Normalized Bit rate (C X R) / ( HX V X F)
Group of pictures (GOP) GOP – Intra (I) frame – Predicted(P) frame – Bidirectional (B) frame I and P – Anchor picture – Stored in memory – Basis for prediction of B picture
GOP in ISO/IEC Coding order != Display order Closed vs. Open Refer handouts for more on MPEG1 and MPEG2
Motion Estimation (ISO/IEC ) Illumination part only 16 X 16 macro block P frame – P predicted from R (P or I) – For each MB in P, find the best match in R – Record motion – Compensate MB – MB’ – No standard for matching algorithm, decision algo.
Motion Estimation(ISO/IEC ) B frame – B predicted from two Rs – Same as P
Rate Control (ISO/IEC ) See whiteboard Constant quality, variable bit rate (VBR) Variable quality, constant bit rate(CBR) CBR rate control
MPEG2 scalability modes Spatial – Base layer – Enhanced Layer – Wavelet? Temporal Adjustable
ISO/IEC Provides very low Bitrate, Error resilience Can be carried in MPEG-2 system layer Entity(Text, graphics, …) can be overlaid on MPEG-2 Video for enhancement content Synthetic audio and video(2, 3D graphics) Support interlaced/progressive scanning
ISO/IEC Goodbye BLOCK OBJECT BASED REPRESENTATION Object description is required Shape coding Content layers are selected at MPEG4 Terminal
ISO/IEC Client can manipulate the contents at his terminal - INTERACTIVITY Server requires back channel Compression rate – Vs. MPEG2 – Discuss how can we increase the compression ratio
Object in ISO/IEC Scene Graph (Video, audio, text and graphic) Transformation and properties are specified Visual object and audial(acoustical) object Compound object can be generated Synthetic and natural object System supports multiplexing and synchronization between media objects
MPEG4 Video coding scheme Shape coding and motion compensation Shape adapted DCT Local block based vs. global (Sprites) Motion compensation First compute static sprite Transmit 8 or 6 global motion parameters Moving foreground is transmitted separately
ISO/MPEG N2728 Multimedia Contents Description Interface Meta data of multimedia data Personal preference, permission Applications – Education, Journalism, Gallery – Entertainment(Karaoke) – Investigation, Surveillance, recognition, GIS, remote sensing – Biomedical, shopping, dating service – Video to text
MPEG11 I GOT YOU !!!
MPEG 21 Description language(MPEG7) & Not for contents but for Media