Progressive Image Coding Based On Mean Block Of Wavelet Coefficients

This paper is mainly devoted to the problem of how the correlation of the adjacent coefficients in the wavelets transform and the multiscale mean values are used in image-layered coding and transporting. First, the wavelet transform is studied. As the development of harmonic analysis, wavelet theory has met wide application dependent on its engineering background, which is very suitable for the research of image coding. The characteristics of energy concentrate and multiscale similarity in the wavelet transform make it possible to attain high performance. Second, some embedded image coding algorithms are studied and compared. EZW (Embedded Zerotree Wavelets) image coder is implemented. Third, on the basis of analyzing of BMMV technique, a new progressive image coding algorithm is proposed. The algorithm creates a good subjective perception of the image at very low bpp, and recognizable images at every scale. Finally, a discussion of the evaluation of the progressive image is made.