AN ADAPTIVE LINER REGRESSION APPROACH TO STILL IMAGE COMPRESSION

摘要

Digital media compression introduces the advantages in storage and transmission. However, the compression procedures also take up computation time. This paper presents an adaptive regression method applied to the standard JPEG compression for archiving higher compression ratios. According to experiments on those popular images, the improvement in compression rate is about 8%, to 18%. Besides, an interesting coding scheme, Golomb-Rice coding, is introduced also for improving the coding and decoding procedures of the JPEG standard. Since the coding scheme is not based on frequency analysis from certain of images to gain a codebook, the decoded images that are encoded with this scheme are assured in average quality. The coding technique can be encoded and decoded in two-directions, since the code-words are symmetric. In other words, the coding techniques have increased encoding and decoding work efficiency two times.