Image compression and transmission using wavelets and vector quantization

To compact a large amount of multimedia data, and route it through a busy network like the Internet, or a low bandwidth one like the telephone network at interactive rates has emerged as a significant contemporary technological challenge. In this dissertation we take a unified “systems view” approach to developing real-time image and video compression/transmission systems. The objective is to produce systems with variable rate/distortion encoding at the source that can flexibly adapt to available channel bandwidth.; There are three aspects to the overall impact of this research in the area of image and video coding. First, our new wavelet coding algorithm is a fundamental contribution in wavelet based compression. It can be used by all wavelet coding methods, and has been demonstrated to improve the performance of the most successful ones in existence. It also led to the first known wavelet coding algorithm with flexible data granularity. Such algorithms can be used in compression/transmission systems that are faster and more reliable than existing ones based on wavelet coding. Second, we have developed the first known compression algorithms that can automatically adapt to fluctuating available bandwidth, while maintaining joint rate distortion optimality. These algorithms are very suitable for use in current compression/transmission systems that use the Internet. Third, we develop a new video coding algorithm, in which we introduce the concept of a re-trainable motion transform. This method is faster and more accurate than older ones like MPEG-1, and of wider applicability than newer ones based on 3-D object modeling. This algorithm may be incorporated into the flexible current MPEG-4 standard for video coding.