Image De-noising And Image Fusion Based On Super Contourlet

Wavelet transform is widely used in signal analysis and processing And achieved a more prominent effect. However in high dimensions, it can't take advantage of the most of geometrical characters of the objects. These geometrical characters are the most aspects we are interested in. People propose multiscale geometric analysis methods on the basis of wavelet transform, The Stationary Wavelet-Based Nonsubsampled Contourlet is a new multidirection and multiscale analysis. It is based on stationary wavelet transform and nonsubsampled directional filter banks. Because of its multidirection and multiscale properties, it can effectively deal with the high dimensions of image. Image denoising and image fusion is the most basic and important elements of the low-level visual in the field of computer vision. The Stationary Wavelet-Based Nonsubsampled Contourlet transformation and its applications on image denoising and image fusion are researched in this dissertation. The main work can be summarized as follows:The Stationary Wavelet-Based Nonsubsampled Contourlet Transform had better performance on ablilty of direction distinguish and shiftinvariant, multiscale, multidirection compared with Contourlet and NSCT. We studied the Stationary Wavelet-Based Nonsubsampled Contourlet Transform coefficients and found that the interscale, interdirection, interlocation dependencies of the its coefficients with respect to signals and noise are quite different. According to this property. we proposed a new algorithm is based on the correlation of the Stationary Wavelet-Based Nonsubsampled Contourlet Transform coefficients effectively. The theoretical analysis and the numeral experiments show that the proposed method can better distinguish the signal and the noise and outperforms the other corresponding contourlet methods in terms of both peak-signal-to-noise ratio (PSNR) and visual quality, and it is especially adequate to the images with much texture.The goal of image fusion is to get clearer image from a special scene so that the image can be further processed and object recognition can be performed better. we proposed a new image fusion algorithm based on the correlation of the Stationary Wavelet-Based Nonsubsampled Contourlet Transform. Local area standard deviation and directional contrast is used to be image features. then its consistency are validated for making fusion result become more smooth. The results of our experiments show that the algorithm can maintain good texture,contour information and a strong contrast, it obtain better visual quality. The Stationary Wavelet-Based Nonsubsampled Contourlet Transform is redundancy and shiftinvariant, so the result effectively suppression the phenomenon of the Gibbs.