| With the rapid development of communication and information technology, the continuous progress of network technology and multimedia technology, images are becoming more and more widely used. In practical applications, images usually have a huge amount of data and take up more storage space and transmission bandwidth, which cause much inconvenience in image-related applications. In this situation, it is especially important to use image compression technology to reduce the storage space and transmission bandwidth. Therefore, the research of image compression is an urgent and realistic subject.At present, the research of still image encoding compression is usually based on the most widely used JPEG international standard. JPEG and its improved algorithms eliminate spatial redundant information in frequency domain, but this approach does not make full use of the characteristic that adjacent pixel may have the same texture, and its effect of spatial redundancy elimination is not very ideal, it has a certain impact on the compression efficiency.In this thesis, a JPEG-like image compression scheme based on H.264 intra prediction is investigated, it combines H.264 intra prediction with JPEG basic framework. According to the characteristic of image texture direction, it can directly eliminate the redundancy of adjacent pixels in spatial domain. On quantitation, it only passes a quantization factor, and avoids passing complex quantization tables like JPEG. It designs the binary code tables for mode compression and compressed file package format. The entropy coding section uses QM arithmetic coding from the JPEG extended system, compared with Huffman coding, it is closer to the limits of information entropy, has higher compression efficiency. Encoding and decoding of 8 bit color and gray still images are realized in language C. When at middle and high bit rate, it can achieve better compression effect than JPEG. |
