The Direction Of The Multi-scale Analysis Of The Application In Image Compression

Along with the development of the information technology, it is demanded to storage, record and transmit kinds of static images and"concretionary"images largely, such as the files, authentic works, images, nephograms and remote sensing images. So it is required that the quality of the image is high using the corresponding channel existent or developing with the stable and credible equipment, but also the costs are low. One of the ways to solve the problem above is actualizing the digital corresponding, transforming the image to the digital signals, wiping off the redundancy outlying of the image quality. Furthermore, on the precondition of ensuring the quality, storaging, recording and transmitting the digital signals is oecumenical, credible and all wool and a yard wide using least bits or bits ratio. Our work in this thesis focus on the research on perceptual image compression algorithm based on transform field and it also belongs to image lossy coding approach.Admittedly, wavelet has a wide application in image processing, especially in image compressing which is one of the most successful application, because of its time-frequency localization and multiscale features. Wavelet, however, is the best base of functions with point singularity, so it can hardly characterize the two-dimensional geometrical structures such as edges and textures in images. Thus, wavelet is not the best base for image sparse presentation. Moreover, the aim of image lossy compression is to eliminate or reduce that visual information with less perceptual importance in order to raise the compression ratio; or make the distortion resulted by quantization to be masked by the image self as much as possible at a given bit ratio in order to enhance the perceptual quality of reconstructed images. Therefore, to overcome the problems discussed above, we make deeply study for the multiscale geometry analysis; build a quantitative model of HVS based on the Contourlet transform; and finally propose our own perceptual image coding algorithm.To begin with this thesis, using Contourlet transform as an example, it conducts the estimation of marginal and joint statistical distribution of Contourlet coefficients via moment method and maximum likelihood estimator. Subsequently, we modify the EZW and SPIHT codec, which is used in the Wavelet transform domain before, to fit in our model, which is in the Contourlet transform domain, by analyzing the character of the Contourlet transform. Finally, we prove that the SPIHT codec in the Contourlet domain is better not only to the EZW codec in the same domain, but also to the same codec in the wavelet domain and JPEG2000 in depicting the veins, which is observed just by naked eyes intuitively.