Two-stage Multiple Description Image Coding Based On Lifting Wavelet

Multiple description coding is a way which can achieve high-quality image signal transmission and video data encoding. It make the signal forming multiple independent stream which have a certain correlation by some ways. When only one description is received, the image which human eye can identify is obtained. With the numbers of received descriptions increasing, the quality of the reconstructed image will be enhanced. It is meaningful to study the construction of multiple description coding. So, we do the following work:We propose a two-stage multiple description image coding based on EZW coding and lifting wavelet. In the program, we propose a two descriptions coding and a four descriptions coding which combine the advantage of the lifting wavelet with two-stage structure. The first stage data is transmitted in the side channel which is the signal after EZW coding. The two stage data is also transmitted in the side channel which is obtained by the signal after lifting wavelet and the first stage signal after EZW decoding. In the decoding side, we consider two cases, including receiving one description and receiving two descriptions. In the case of four descriptions, We propose a progressive interpolation algorithm. Through the experiment, we find our algorithm has good ability to resist channel noise, and gets high quality reconstructing image signal.We propose a two-stage multiple description image coding based on SPIHT coding and lifting wavelet. We study the cases of two descriptions and four descriptions. We also propose a new method to the case of two descriptions, three descriptions and four descriptions in the four descriptions algorithm, which use different threshold for the pixels of the most low frequency to image data fusion. Through the experiment, our algorithm can save a lot of decoding time, improve efficiency greatly.