Signal compression for digital television.

Abstract

Still image and image sequence compression plays an important role in the development of digital television. Although various still image and image sequence compression algorithms have already been developed, it is very difficult for them to achieve both compression performance and coding efficiency simultaneously due to the complexity of the compression process itself. As a results, improvements in the forms of hybrid coding, coding procedure refinement, new algorithms and even new coding concepts have been constantly tried, some offering very encouraging results.In this thesis, Block Adaptive Classified Vector Quantisation (BACVQ) has been developed as an alternative algorithm for still image compression. It is found that BACVQ achieves good compression performance and coding efficiency by combining variable block-size coding and classified VQ. Its performance is further enhanced by adopting both spatial and transform domain criteria for the image block segmentation and classification process. Alternative algorithms have also been developed to accelerate normal codebook searching operation and to determine the optimal sizes of classified VQ sub-codebooks.For image sequence compression, an adaptive spatial/temporal compression algorithm has been developed which divides an image sequence into smaller groups of pictures (GOP) using adaptive scene segmentation before BACVQ and variable block-size motion compensated predictive coding are applied to the intraframe and interframe coding processes. It is found the application of the proposed adaptive scene segmentation algorithm, an alternative motion estimation strategy and a new progressive motion estimation algorithm enables the performance and efficiency of the compression process to be improved even further.Apart from improving still image and image sequence compression algorithms, the application of parallel processing to image sequence compression is also investigated. Parallel image compression offers a more effective approach than the sequential counterparts to accelerate the compression process and bring it closer to real-time operation. In this study, a small scale parallel digital signal processing platform has been constructed for supporting parallel image sequence compression operation. It consists of a 486DX33 IBM/PC serving as a master processor and two DSP (PC-32) cards as parallel processors. Because of the independent processing and spatial arrangement natures of most image processing operations, an effective parallel image sequence compression algorithm has been developed on the proposed parallel processing platform to significantly reduce the processing time of the proposed parallel image compression algorithms.