Vector Quantization Technique And Application With Parallel Program

In recent years, parallel and distributed processing have been conjectured to be the most promising solution to the computing requirements of the future. Significant advances in parallel algorithms and architectures have demonstrated the potential for applying concurrentcomputation techniques to a wide variety of problems.With the continuing growth of the World Wild Web services over theInternet, the demands of rapid image transmission over a network link of limited bandwidth and economical storage of a large image database are increasing rapidly. One of the emerging technologies for lossy image compression is Vector Quantization (VQ). Vector Quantization can be used to take advantage of the correlation between neighboring pixels by quantizing pixels in groups (or vectors) rather than individually and symbolically representing the vector with a codeword.We start with the theory of Vector Quantization, and come on with the application of of V.Q. technics on image processing. Fist we developed a Codebook creation algorithm based on the Partial distortion theorem, which has greatly improved the quality of the initial codebook. Then we studied the classific V.Q. and got a novel CVQ algorithm. Because the data of CVQ has been split by the algorithm already, it is the right occasion of of parallel program with domain partition. So we implement the parallel algorithm of this algorithm. Finally , we discussed some problems which is the barrier of the application of V.Q. technic.