Research On Image Compression Based On Wavelet Transform

With the ever-growing multimedia technique, people are looking forward to new image compression techniques with better performances. Taking the storage and transfer of mass image data into consideration, not only could favorable image compression techniques achieve high compression ratio and fidelity, they should also meet the demands of web applications such as progression transmission. Thereby, to satisfy the needs of practical applications, it is necessary to research on image compression further in detail.The paper discussed the wavelet theory firstly, and then EZW and SPIHT algorithms are discussed and analyzed in detail. Although SPIHT provides flexible quality layered embedded bit-stream, it doesn't support resolution scalability. By going deep into wavelet transform and SPIHT algorithm, an improved SPIHT algorithm has been presented, and has been proved not only to be able to keep the same compression efficiency as SPIHT method but also support resolution scalability. When people don't need high resolution reconstructed images, they can get much higher quality reconstructed images than SPIHT by decoding proper low resolution images using improved SPIHT algorithm. Thirdly the general information of the JPEG2000 still image compression standard is outlined, and each part of the JPEG2000 coder is discussed in detail according to the JPEG2000 encoding order, especially about the EBCOT codec. Finally the ROI coding methods in JPEG2000 and PSBShift method are then discussed and analyzed in detail, and an improved PSBShift is proposed. Then we realized Maxshift method and improved ROI coding method. The experimental results show that the improved ROI coding method is much more accordant with HVS and suitable to progressive image transmission.