Research On High Compression Ratio Static Image Coding Based On Human Visual Characteristics

DCT transform based on the block was often used as core technology in current international standard of image encoding. The main characteristics of encoding based block transform is block effect will appear obvious in high compression ratio,which reduce the subjective image quality. In order to reduce the blocking effect under the condition of high compression ratio, many different solutions has been proposed, and this is also what we are going to study. In order to reduce the block effect in flat areas and retain more image edge details, we propose an compression algorithm based on adaptive block classification by HVS in DCT domain, which effectively improve the compression quality.In this Thesis, Firstly core transformation was systematic analyzed based on learning algorithm of image compression and encoding standard, and compress JPEG, JPEG2000 and JPEG-XR, by which analysis their compression effect on high compression ratio; And then, mainly analysis the characteristics of human visual system and on this basis, a new idea of block classification of image compression based on visual characteristics is proposed, and also a new method of image block classification on DCT domain is suggested. Which by calculating the activity of the image block, the image block is divided into smooth area, edge area, texture area three categories; finally, take an analysis of distortion quantization by unified quantization table, and find the fundamental reasons of block effect on JPEG, then flat area, edge area and texture region are adaptive quantization combined with the characteristic of human vision. Adaptive quantization compressed image has better visual effect and compression quality.