Overview	Outline	Instructors	Online Registration	About Portland

Overview

The past decade has witnessed a great surge in the need for storage and transmission of digital images and video. Digital still cameras, camcorders and inexpensive scanners have proliferated the use of digital images in various consumer and commercial applications. Also, the widespread use of the Internet and multimedia enabled PC's, the advent of digital broadcast systems such as DSS and HDTV, and the affordable storage of digital video on CD-ROM and DVD have made digital video an integral part of everyday life. Despite the continuing increase in capacity, efficient storage and transmission of images and video is still the foremost challenge in all those systems. Consequently, image and video compression plays a key enabling role for many consumer, commercial, and scientific applications.

Digital image and video compression is a current focus of both research and international standardization. Recently developed standards such as JPEG, H.261, MPEG-1, MPEG-2, H.263 (and its H.263+ and H.263++ enhancements), and MPEG-4, and emerging standardization projects such as JPEG2000 and H.26L reflect the state of the art in visual content coding techniques. They are an important factor for facilitating interoperability among various imaging systems as well as for deploying the technology widely and in a cost-effective manner.

A recent MPEG standardization activity that has started early in 1999, MPEG-7, is addressing standardization of multimedia content interfaces. MPEG-7's scope includes standardization of descriptions of multimedia information, as well as descriptions of user preferences and usage history, for filtering, browsing, searching and retrieval applications. MPEG-7 responds to the need for managing digital audiovisual information that is becoming increasingly available to users through digital TV broadcast, digital video cameras, DVD, and PC-based access to multimedia on the Internet. Further, persistent storage that allows non-linear access to audiovisual material, such as HD storage in PC platforms and personal video recorders (PVR), is becoming available in consumer devices. Consequently, there is need for rapid navigation and browsing capabilities to enable users to efficiently discover and consume the contents of audiovisual material. MPEG-7 will attain an international standard status at the end of year 2001.

This course provides the audience with a solid understanding of the fundamentals, as well as a working knowledge of the various image and video compression standards and the applications that they serve. Upon completion of the course, the students will be equipped with the fundamental knowledge that will help them understand the performance limitations of various compression solutions. Further, they will gain a solid understanding of the principles of the various digital image and video compression standards in such a way as to optimize their use for a particular application. Product demonstrations of standard compression algorithms in database applications, prerecorded DVD media, PC-based video teleconferencing systems and Internet streaming media systems will be furnished.

The introductory section of the course starts with a broad set of product and application examples that establish the need for compression followed by a brief description of the existing and emerging image and video compression standards that serve these products. The main part of this section is a detailed description of the four fundamental building blocks of compression systems, namely, motion compensation (with fixed and variable block sizes, conventional and overlapped blocks, integer and fractional motion vector accuracy, etc.), transformation (Discrete Cosine Transform, wavelets), quantization (SQ, UTQ, etc.), and symbol modeling and encoding (Huffman, 2-D and 3-D run-length and level coding, and context-based arithmetic coding).

The main focus of this course is on image and video compression standards. First, a detailed description of the lossy JPEG still-image standard, which is also the foundation for video compression schemes such as the MPEG and the H.26x family of standards, is provided. Various issues regarding the implementation of JPEG in practical systems are discussed. These include encoder optimization, e.g., the design of Huffman and quantization tables and rate-distortion optimized JPEG; decoder optimization, e.g., blocking artifact removal and post-processing of JPEG compressed images; JPEG enhancements; and other implementation issues such as bitrate/image quality tradeoffs, fixed-rate JPEG, the effect of multiple coding and software and hardware speeds. Numerous image examples supplement the technical descriptions.

The ISO JPEG committee is currently developing a new still-image compression standard in seven parts, referred to as JPEG-2000, which is based upon wavelet decomposition. Combined with powerful quantization and encoding strategies such as embedded quantization and context based arithmetic coding, the use of wavelets in JPEG-2000 provides the potential for numerous advantages over the existing JPEG standard. Performance gains include improved compression efficiency at low bit rates or for large images, while new functionalities include multi-resolution representation, SNR scalability and embedded bit stream architecture, lossy to lossless progression, region-of-interest (ROI) coding, improved error resilience, idempotency to multiple compression cycles, and a rich file format. In this course, practical implementations of the wavelet transform, as applied to image compression (e.g., memory efficient implementations and the lifting scheme, various integer and floating point bi-orthogonal filters for lossless and lossy compression, etc.), and related quantization and coding strategies are discussed. The technical details of the JPEG-2000 Part 1 algorithm (which became a DIS in December of 2000) and syntax are reviewed extensively and the JPEG-2000 performance is compared to that of the existing DCT-based lossy JPEG standard. The superior features of the JPEG-2000 proposed standard are demonstrated by numerous image examples.

The remainder of the course is devoted to video compression algorithms and standards. The major components of a video compression system, namely motion estimation and compensation, pre-processing, encoding, decoding, and post-processing are discussed. First, fundamental principles of motion estimation and a detailed discussion of widely used block based motion estimation algorithms are provided. The applications of motion estimation in multi-frame pre-processing algorithms are reviewed next, followed by a review of algorithms for motion compensated noise suppression and defect removal. Pre-processing is often among the differentiating factors used in evaluation of an entire video compression system due to its significant impact on the resulting compression efficiency.

The final portion of the course is devoted to the video compression standards, namely, the MPEG-1, MPEG-2, H.261, H.263, H.263+, H.263++, MPEG-4, and "H.26L" standards. A detailed description of MPEG-1 is given followed by a presentation of the MPEG-2 video compression standard with an emphasis on its differences from MPEG-1. An overview of MPEG-2 systems is provided with an emphasis on the applications of MPEG-2 transport for carriage of streams auxiliary to audiovisual streams, namely program and service information and other types of data. Its applications in DVD and DTV standardization are also discussed. A detailed overview of the recent MPEG-4 standard is presented with an emphasis on major differences between MPEG-4 and preceding MPEG standards. MPEG-4 introduces object based coding, coding of synthetic and natural objects, and object based scene composition and manipulation. Fundamental aspects of object based coding, such as coding of arbitrarily shaped video objects, will be presented. An overview of MPEG-4 Systems will be given to demonstrate object based scene composition and manipulation functionality of MPEG-4. MPEG-4 targets new application areas with increased interactivity and extendibility. MPEG-4 completed its first version in 1999 and has since completed three amendments to its visual standard to add more new features.

Next, the ITU video coding standards are reviewed beginning with H.261, the first practical digital video coding standard. H.261 is still widely used for backward compatibility in video conferencing scenarios (e.g., ISDN use at 128 kbits/sec). The H.263 standard is then discussed, including its recent H.263+ and H.263++ enhancement projects. H.263 has provided a significant advance in coding efficiency over previous technology, and was used to form the efficient backbone of the MPEG-4 design. It was the first standard to demonstrate acceptable video quality over PSTN networks using 28.8 kbits/sec modems and brought video conferencing to the mass consumer market as well as displacing H.261 as the primary video conferencing standard for all bit rates. The H.263+ project, completed in early 1998, enhanced the performance of H.263 and extended the standard to a broader range of applications, including enabling the first highly effective error resilient standard video coding for mobile corrupting channels and for Internet packet video. The H.263+ standard incorporates numerous enhancements and new features over the original H.263 design and is even better suited to low data rate and Internet applications. The H.263++ project, just completed in 2000, extended those capabilities even further.

The "H.26L" next-generation standard development project will then be discussed, along with projections of what quality improvements this standard will bring. The work conducted toward the development of this new standard promises significant compression gains over any previous standard design. It also includes a new structure for video coding development, separating the codec design into two distinct layers: a video coding layer (VCL) to efficiently represent the video contend, and a network adaptation layer (NAL) to customize and packetize the content for delivery over a particular network environment.

Next, topics in Internet video will be presented, including a network infrastructure review, streaming video examples and discussions on Internet packet loss and error resiliency techniques. Included in this portion of the course will be numerous demonstrations of the individual video compression standards and systems. Finally, we will present an overview of the scope and applications of the emerging MPEG-7 standard where we will also introduce the fundamental concepts in MPEG-7. We will demonstrate a prototype MPEG-7 player, illustrating the personalization, browsing, and navigation functionalities enabled by MPEG-7.

Return To Index