Research For Region-of-Interest Coding Based On Wavelet And Contourlet

With the development of network technology and the popularity of multimedia technology, image storage and transmission become an essential step since image is an important factor in the multimedia technical information. An image with a large amount of data is stored for more effective and reducing the problems in the communication process, so we aim at seeking a good image compression coding technology.Under the situation of demanding more advanced image, image with a region of interest (ROI) compression coding emerged. In JPEG2000 standard for the ROI coding proposed two basic encoding methods, the general scaling based method and the maxshift method. Many studies on the region of interest coding method are based on the above two.The main contributions of this dissertation are recapitulated as follows:1. For the ROI image compression, wavelet transform has received success. A large number of experiments confirm that the wavelet coefficients are the importance of the whole image. The experiment upgrades the wavelet decomposition low-frequency and partial high-frequency coefficients by the general scaling method of JPEG 2000, in order to improve the quality of interested regions and the whole images, and then code combining with SPIHT. Through three sets of contrast experiments, we can reach this conclusion: reconstructed image quality has been increased after completely improving low-frequency and intermediate frequency coefficients.2. For the texture-rich of ROI image, this paper presents an effective compression scheme, namely a region of interest based on class WBCT image compression. Firstly, we decompose the image with combining wavelet transform and the direction filter in contourlet transform. Secondly, we use the maxshift method to upgrade the ROI coefficients in the wavelet domain. Finally, coding is implemented with SPIHT algorithm. After coding for large amount of image data,the results show that this scheme has good invisibility to the ROI image with the rich texture.Wavelet transform can not effectively deal with high dimensional data for the texture-rich of images. Compared with the wavelet transform, contourlet transform not only has good directionality and anisotropy, but also can efficiently capture the image geometry.Some basic research related with image compression based on ROI image compression has been done in this paper, but many problems still need to be analyzed and researched deeply, for example, how to choose the ROI better, how to choose the coding compression better.