Research On Image Compression Technology Based On Wavelet Transform

The technique of multimedia communication develops rapidly with the increasement of application demands. Image is one of the most important medium in the multimedia field, but the amount of data is so large that it is necessary to be compressed due to the difficulty for storage and transmission. Wavelet transform can catch the nonstationary information of image with the local time-frequency property, which can improve compression ratio. Furthermore, the transform scheme is consistent with the characteristic of human eyes, which provides a capability of multilevel encoding and avoids the dispersion of encoding error effectively.According to the appilication of wavelet transform in image compression, this thesis puts focus mainly on the folowing subjects:Firstly, this thesis intruduced the basic theories and methods related to the object, including image compression, Wavelet Transform, and the Embedded Zerotree Wavelet algorithm, EZW, etc.And then, the main content of the thesis is descripted in detail. An overview of the EZW algorithm is given, its defeat is analyzed, and the way to improving its compression ratio is pointed out and an improved algorithm is investigated.Again, a new method of video coding based on the improved three-dimension wavelet transform (3D-WT) and motion compensation is proposed according to the optimal bit allocation strategy. With the new algorithm, video sequence is firstly separated into two groups according to the parity of each frame. Then temporal decomposition is applied to each group separately, the results including several high frequency frames and a low frequency frame in each group. After that, the motion compensation is only applied in the two low frequency frames in order to eliminate the interframe redundancies. Experimental results show that the proposed scheme has better coding performance compared with other schemes based on 3D-WT.The conclusions are made in the last chapter, and some research topics along the lines of this thesis in the future and the possible solution idea are pointed out.