Research On Embedded Wavelet-Based Image Coding Algorithm

Image compression can be realized by removing all kinds of redundancy in image. Classical image coding technology mainly based on the inherent statistical property of image itself. Modern image coding method does not limit to the information theory frame, by making use of human visual property and the characteristic in image, it can obtain higher compression ratio and gets better subject visual quality of recovery image.The wavelet analysis theory was the powerful tool of non-steady signal analysis. Wavelet-based image coding have been widespread applied in image compression for having better rate distortion characteristic in low bit rate and can produce embedded bit stream of quality-graded, it has already became important part of international standards (JPEG2000) for image compression. Lifting wavelet is the key technology to construct second-generation wavelet. Comparing with traditional wavelet transform, it does not rely on Fourier transform, being easy for computation, have low complex in time domain and spatial domain ,and is easy to realize for wavelet-based image coding.Wavelet-based EZW,SPIHT algorithm have important influence to the present wavelet-based image coding for having good compression performance and being easy to realize. But they have not been able to make full use of distribution characteristic of wavelet coefficient, coding efficiency will decrease along with the increasing of compression ratio. On the other hand, not considering the human visual system (HVS) will affect the quality of recovery image. How to get good visual quality under high compression ratio is an important research point. The sorting pass extremely affect the efficiency of entire coding algorithm. The thesis research on embedded wavelet-based image coding, mainly analysis the SPIHT algorithm, in view of the shortcoming of the algorithm, proposed an modified algorithm, and analyzed the algorithm performance theoretically .The modified algorithm uses lifting wavelet, based on human visual system, and optimizes sorting pass of SPIHT.The modified algorithm has merit: lifted wavelet transform is used to enhance coding speed and is easy to realize; wavelet coefficients in each subband are weighted by different value to make sure transmit most important coefficient for visual first; making full use of distribution characteristic of wavelet coefficient can remove more redundancy in bit stream and enhances coding efficiency. The simulation result shows that the modified algorithm enhances compression performance and also has good recovery image quality.