Research On Image Compression Algorithm Based On Wavelet Transform

Wavelet analysis theory is a fast-growing emerging subject in applied mathematics and engineering field. Wavelet transform has the character of dual localization in time domain and frequency domain, which benefits the time and frequency domain analysis of signal. As a consequence, wavelet transform has been widely researched and used in image compression. Compared to traditional image compression, Wavelet transform has a more compression efficiency. And it realizes the progressive transmission of compressed signal. The Multi Resolution Analysis character of wavelet transform provides the mechanism of Human visual. The Image Data can maintain the original fine structures under every resolutions, which creates great convenience to delete extra information in image.This paper mainly focuses on approach of the image compression based on wavelet transform, analyses the development and the overview of image compression, introduces the basic theory, methods, international standards and quality evaluation standards of image compression. A further research has been made on foundations of mathematics of wavelet analysis and the theories related with image compression. Optional questions of wavelet base have also been discussed in this paper, which mainly focus on embedded image compression algorithm based on zero tree, including Embedded Zero Tree (EZW) and Set Partitioning In Hierarchical Tree (SPIHT). The principle, process, advantage and disadvantage of the algorithm have been illustrated. Finally the simulation has been made on the two algorithms and a improved EZW algorithm has been proposed according to the disadvantages of EZW algorithm. This new algorithm has made five improvements: 1.visibility preaccentua combined with human visual characterstic, give different weights to different sub-band wavelet coefficients. 2. The lowest frequency sub-band includes most energy of original image, so use DPCM coding to code on the lowest frequency sub-band. 3. Make improvement on threshold quantification. 4. Make improvement on scanning sequence of zero tree. 5. Make improvement on Successive Approximation Quantization(SAQ), reduce the time of approximation quantization, accelerate the coding speed. Compared with the improved EZW algorithm to the traditional EZW algorithm, it is show that the new algorithm has better effects both on objective compression performance and subjective visual feeling of reconstructed image.