This research involves the development of efficient, reliable, and real-time realizable coding algorithms and techniques for compressing and transporting audio-visual signals, targeting particularly the visual-telephone application. For a total channel bit-rate of 64 kb/s or less, our goal is to achieve superior grade-of-service by treating the compression of the audio (speech) and video signals as an integrated problem. Our thrust centers around the use of vector quantization techniques, and schemes that jointly optimize the signal analysis and quantization functions. We place particular emphasis on finding an efficient representation of the displacement ("motion") field for motion compensated prediction of the video signal. To facilitate personal wireless communications, we are exploring adaptive coding schemes that exploit the prioritized transmission of encoded data. For compressing speech signals, we are exploring variable rate coding schemes, and are allocating bits between the speech and video signals to optimize the overall grade-of-service. Achieving our goal would give significant impetus to the development of the emergent Information Superhighway, which can form part of an infrastructure for supporting high performance computing. Educational activities undertaken in conjunction with the above research include but are not exclusive to: the development of undergraduate digital signal processing laboratory courses; the introduction of graduate courses in the area of audio-visual information processing and communications; increasing the use of computer-based multimedia instruction in the classroom; assisting in speeding up the digitization of the library and networking of its databases.
