An Adaptive Partitioned Algorithm Based on Mean Coding and Vector Quantization

A hybrid algorithm based on mean coding and vector quantization is presented in this paper. The smooth image block is coded with the mean coding algorithm, which can reduce the vector quantization (VQ) search computation and improve the compression ratio effectively, and this method is called the basic blended algorithm (BBA). On this base, an adaptive partitioned algorithm (APA) algorithm based on 16×16 pixels image block is proposed, the optional image block sizes are 16×8, 8×16, 8×8, 8×4, 4×8, 4×4 pixels, the image is coded adaptively at different levels. The size of mean coding is chosen adaptively according to the size of the smooth image block, the Not-smooth 4×4 pixels image block is coded by VQ. In addition, the improved APA (IAPA) is also given in this paper. Simulations show that the APA and IAPA proposed in this paper are proven to have great performance for different kinds of images, especially significant for those have large areas of static background and little details. Compared with BBA, the compression ratio of APA and IAPA can be improved by 416% and 512%, the PSNR of APA and IAPA is reduced only 2.4% and 3.6%, the complexity can be reduced significantly, which is of advantage to hardware implementation.