Advanced wavelet image and video coding strategies for multimedia communications.

摘要

Multimedia communication will play an increasingly important role in tomorrow's information infrastructure. Despite the significant effort in both academia and industry for multimedia applications, which resulted in a large number of standards and commercial products, providing high quality multimedia communications over either wireline or wireless networks still represents a major challenge. In the dissertation, several image and video coding techniques are developed enabling highly efficient visual communications over both wireline and wireless networks. Finally, a complete mobile wireless visual communication system is presented integrating our video coding technology with networking infrastructure.; We start by describing our high performance wavelet image codec developed in the Multimedia Communications and Visualization Laboratory. Significance-linked connected component analysis (SLCCA) exploits both the within- subband clustering property and cross-scale dependency of wavelet coefficients. As a result, SLCCA is among the best wavelet codecs reported in the literature.; For video technology, three techniques are proposed, namely, video significance-linked connected component analysis (VSLCCA), three-dimensional significance-linked connected component analysis (3D-SLCCA), and Haar significance-linked connected component analysis (H-SLCCA). In VSLCCA, fine-tuned motion estimation is developed to reduce temporal redundancy and exhaustive overlapped block motion compensation is utilized to ensure the coherency in motion-compensated error frames. Thus wavelet transform and the following SLCCA strategy can be efficiently applied rendering superior performance to state-of-the-art H.263 standard. In 3D-SLCCA, both the wavelet transform and SLCCA data organization and representation strategy are extended to three-dimensions to include the time domain. 3D-SLCCA provides scalability, symmetric low computational complexity, and moderate robustness against transmission error propagation thus being well suited for wireline video. H-SLCCA was specifically developed for mobile video communications. In addition to its low complexity, H-SLCCA is highly error resilient, preventing both spatial and temporal error propagation.; In a real communication scenario, in addition to source coding, transmission error protection must also be considered. The SLCCA codec is extended in this dissertation by using joint source-channel coding strategies. The integrated error resilient tools (error resilient packetization, synchronization, and unequal forward error correction) extensively exploit the properties of the SLCCA bitstream. For a binary symmetric channel, the resulting codec outperforms Sherwood and Zeger's technique, who reported the best results in the literature.; Finally, we propose a complete wireless video communication system using application driven and integrated system design. For efficient visual communications, the application layer is responsible for video coding, transmission error protection, and packetization as well. The developed system is able to provide high quality video over low bandwidth error-prone wireless links.