Abstract: First the PDVQ（Partition Dynamically VQ） image encoding system categorizes the codebook based on the directions of the codewords，in each category codebook all codewords are sorted in the ascending order of their sum，and then each category codebook is divided into blocks in average.During the encoding process，the encoding system analyzes the correlation of the input image block and its neighboring blocks using PDVQ algorithm，and judges the direction of input image block to choose corresponding category codebook.And a nearest codeword to the input image block can be found out in the search range which can be gotten based on the sum of the input image block using EBNNS（Equal-sum Block-extending Nearest Neighbor Search） algorithm.Experimental results demonstrate that this encoding system integrates the merits of PDVQ with EBNNS and category encoding algorithm based on direction，which can improve the compression rate and coding speed in contrast with the normal VQ，and will not affect the quality of the image reconstructed.