ECE 617 - Multimedia Data Compression: Theory, Algorithms, and Applications
Instructor
Professor En-Hui Yang
Calendar description
Overview of multimedia communications system, digital representation of multimedia signals, introduction to multimedia coding theory, entropy, rate distortion function, Huffman coding, arithmetic coding, run-length coding, Lempel-Ziv coding, quantization, Lloyd-Max algorithm, JPEG compression, hybrid video coding, MPEG 4 and H.264 coding standards, transmission protocols (TCP/IP, RTP, RTSP) and channel modeling, error detection and erasure recovery, error concealment, multimedia security, watermarking.
This course is held with ECE 415. In addition to the above contents, graduate students will have opportunities to learn the following advanced materials through research projects and/or supervised reading as well: Kolmogorov complexity, grammar-based coding, search- able compression, graph and tree compression, interaction between data compression and big data analytics, compression for cloud storage, computational approach to lossy compression, soft decision quantization, and rate distortion optimization in JPEG, H.264, and HEVC.
Prerequisites
Linear algebra, advanced calculus, and probability and statistics.
Textbook
Complete course notes plus papers and coding standards will be available on UW Learn.
Reference books
- P. Havaldar and G. Medioni, Multimedia Systems: Algorithms, Standards, and Industry Practices. Course Technology, Cengage Learning, 2010. (ISBN-13: 978-1-4188-3594-1)
- Iain E.G. Richardson, H.264 and MPEG-4 Video Compression. New York, NY: Wiley, 2003.
- W. B. Pennebaker and J. L. Mitchell, JPEG: Still Image Data Compression Standard. Norwell, Massachusetts: Kluwer Academic Publishers, 2003
Detailed description
- Introduction: overview of multimedia communications system and requirements; digital representation of image and video, image aspect ratio, color space and linear color transform, and YCrCb sampling formats; quality measurement; multiple resolution and description representation by source coding; multimedia security and watermarking; random sequence. (Lecture hours: 3)
- Fundamentals of lossless source coding: entropy, joint entropy, and conditional entropy; variable length codes and Kraft-McMillan inquality; source coding theorem; Huffman coding; universal source coding; run-length coding; arithmetic coding; Lempel-Ziv coding. (Lecture hours: 10)
- Introduction to lossy source coding: mutual information; information quantities for continuous random variables; rate distortion function; Gaussian rate distortion formula; quantization; Lloy-Max algorithm for designing optimal scalar quantizers. (Lecture hours: 4)
- Image and video coding: basic ideas of predictive coding and transform coding; Karhunen-Loeve (KL) transform; discrete cosine transform (DCT); JPEG standard and sequential DCT-based mode of operation; examples of multiple resolution coding: progressive DCT-based and hierarchical modes of JPEG; rate distortion optimization in JPEG; hybrid video coding; temporal prediction and motion compensation; overview of H.264 standard; the baseline profile of H.264 codec. (Lecture hours: 12)
- Transmission and recovery: Transmission protocols (OSI architecture, TCP/IP, UDP, RTP, RTSP); channel modeling; channel capacity; linear block codes; error detection; erasure recovery; error concealment. (Lecture hours: 7)
- Advanced materials: Kolmogorov complexity; grammar-based coding; searchable compression; graph and tree compression; interaction between data compression and big data analytics; compression for cloud storage; computational approach to lossy compression; soft decision quantization; rate distortion optimization in H.264 and HEVC; distributed coding; and multiple description coding.