Scalable video for multimedia workstations

Abstract This paper provides an overview about the problems of motion video representation for multimedia workstations and reviews the state-of-the-art. In addition to highly efficient data compression, coding algorithms for motion video have to provide extra flexibility to accommodate the varying computational, bandwidth and display parameters of open computer systems. This “scalability” should include image size scalability, partial bit-stream decodability, and computation-limited coding and decoding. Techniques standardized to date, such as CCITT H.261 or MPEG-1, are not ideally suited for motion video integration into computer systems. We show that resolution hierarchies are a straightforward approach to build a scalable video codec. Resolution pyramids offer varying degrees of asymmetry by increasing the coder complexity for a given decoder. Subband pyramids achieve spatial scalability without introducing redundant samples, however, there is no extension to a spatiotemporal system with motion compensation yet. Resolution pyramids with redundant samples can be extended to a 3D spatiotemporal resolution pyramid with motion compensation. Various examples of scalable image representations are given in the paper.