Dynamic window search and probability dominated algorithm for fast vector quantization encoding

Summary form only given. Vector quantization (VQ), which has been widely applied in speech and video coding, provides an efficient technique for signal compression due to its excellent rate-distortion performance. A fast encoding method to speed up the closest codeword algorithm was proposed by accounting for the occurrence of each codeword for every input vector before practical encoding. Simulation results demonstrate that the little processing and memory cost significantly reduces the search time while marinating the same encoding quality as that of the full search algorithm. It was found that the proposed algorithm outperforms all existing VQ encoding algorithms.