Image Coding Basing On Wavelet Transform

In recent years,because of the popularization of computer and internet, as the outcome of computer, video, audio and communication technology, multimedia has became an indispensable part of human ordinary life, and has also became one of the pop technology in computer field. As the rapid development of multimedia technology and communication technology, multimedia entertainment, multimedia communication, digital audio frequency broadcast and high definition television etc, these practical applications enhance the requirement of information data storage and transmission. One of the main problems, which confronted with multimedia technology, the applications and developments of Internet, is to solve the denotation, transmission and storage of the image data information. In order to obtain satisfying vision and transmission effect, effective image data compress coding, is the key technology to solve. The purpose of digital image compress coding is to token the image within the limit of bit, and to maintain the quality of renew image.Firstly, we introduce the development of image compression, the method of image compression, the metewand of image compression performance and the international standard of image compression. Otherwise, we fully discuss and analyze the theory of wavelet transform, and study the approach of image coding through wavelet transform.Secondly, because wavelet transform can realize time-frequency localization and multi-resolution, it has been the main aspect of current image compress coding research. Zerotree makes use of the relativity of child band, which is a meaningful breakthrough based on wavelet image compress coding. EZW and SPIHT algorithms based on zerotree structure, which enhance the image compress ratio largely, are discussed and analyzed in detail.Finally, in order to make the most of the characteristic of wavelet transform, we improve EZW and SPIHT algorithm in different aspect. Firstly, by changing the scan route, some repeated scans are avoided and the encoding time decreases. Secondly, as coding with EZW,'P'and'N'don't contain the information of the subsequence's significant. If the subsequence isn't significant, much sequential'T'appears. Furthermore,'T'contains the information of the subsequence's significant, and the leaf-nodes haven't subsequence, so'T'includes redundancy. In this paper, we propose a scheme based on EZW to reduce'T'redundancy. Thirdly, SPIHT with multi-threshold and feedback of node, reduces the bits which to denote the significance of subsequence and the position redundancy. In the end, the Bin-zerotree-structure, together with feedback of node, can reduce the redundancy of brother and position, so it improves the compression ratio of EZW. Experimental result proves the effect of the improved algorithms.