Embedded video subband coding with 3-D set partitioning in hierarchical trees

This thesis presents a three-dimensional (3-D) embedded video coding algorithm that utilizes 3-D wavelet property and 3-D extension of the set partitioning in hierarchical trees (3-D SPIHT). An embedded representation of video signal, in which a single compressed bit-stream of a video signal provides the different subsets of it, corresponding to the compressed versions of the same video signal at any lower rates, has important applications such as multi-user scenario and video database browsing. In order to achieve the embeddedness, we introduce a novel idea of spatio-temporal orientation trees. By this means, the significance information of 3-D wavelet transformed video is encoded to reduce the mean-squared-error distortion measure. Specifically, by successively partitioning a set of coefficients into its subsets and transmitting significant information first, the algorithm gradually approximates original video with a fine granularity. Extension to color-embedded video coding is accomplished without explicit bit-allocation, and can be used for any color plane representation. In addition to being rate scalable, the proposed video coder allows multiresolutional scalability in encoding and decoding in both time and space with a single bit-stream to adapt to different display parameters. These attributes of full embeddedness and scalability, lacking in MPEG-2 and H.263, along with many desirable features such as precise rate control and low-complexity make the video coder an attractive candidate for multi-media applications.;The proposed video coder was implemented in C++ and was applied to several representative video sequences at a wide range of rates from 30 Kbits/sec to 2.53 Mbits/sec to evaluate the performance in terms of subjective and objective quality of the decoded video, functionalities, and system complexity. Additionally, we investigated a channel error correction technique to protect the 3-D SPIHT bit-stream from channel noise by employing joint source and channel coding method, where we trade off between source coding rate and channel coding rate at a known channel bit-error rate.