Efficient image data compression algorithms and robust image transmission over noisy channels

We primarily focus on two important aspects for image communications, efficient image compression based on wavelet transform and image transmission over noisy channels in this research. One of the major objectives of the research is the development of efficient image compression techniques involving the establishment of effective models and the optimization of the rate distortion performance. We develop a novel predictive tree structure with good performance and less computational complex, including both the optimal hierarchical quadtree construction and the predictive spatial orientation tree development. Regarding the progressive image coding, we present a new embedded image coding scheme using multistage vector quantization based on rate distortion optimization. The classified multistage vector quantization system is used to implement the partially ordering and successive approximation of wavelet coefficient vectors in terms of a set of decreasing rate distortion slopes. The rate distortion slopes of coding vectors are estimated using a multiple context model along with a priori information of the vector quantization system.; The other important objective in this research is the image transmission over noisy channels. We propose a robust image coding system which combines classification and optimal bit allocation along with robust quantization. In order to improve coding efficiency, an iterative algorithm is developed for implementing optimal bit allocation involving both rate distortion and channel optimization. The system achieves both efficient image compression and robust image transmission without channel coding. Joint source and channel coding is also addressed in this dissertation. We propose an efficient joint source and channel coding system for image transmission over additive white Gaussian channels. The source and channel coding are performed in parallel so that the overall bit budget can be distributed optimally between the source coding and channel coding without any knowledge of the image data. We introduce a block based quadtree wavelet image encoder to achieve efficient image coding and avoid error propagation. Various code rates based on turbo codes are employ to implement unequal error protection to the substream according to the importance associated with each coding plane. The system demonstrates the competitive performance in the field.