Remote Sensing Image Denoising Based On NSCT

Remote sensing images in the transfer processing will inevitably be affected by noise, which causes the picture quality to drop and seriously affect the remote sensing images of the analysis and understanding. Therefore, to improve the image quality and outstanding features of the image, image denoising is a fundamental problem. In recent years, the multi-resolution analysis's thought obtained the widespread application in the imagery processing domain and has obtained the huge success. Cunha and M.N.Do developed a fully shift invariant, multidirection, and multiscale transform -the nonsubsampled contourlet transform, which has a fast implementation in image enhancement and denoising.This paper proposes some modified methods for image denoising based on the study of the Nonsubsampled Contourlet Transform, specific tasks are as follows:1. A new local adaptive threshold estimation method for image denoising based on the Nonsubsampled Contourlet Transform is proposed. The new method determines the shrinkage threshold according to the neighbouring NCST coefficients, the scale of the coefficients and the noise level. The simulation experimental results show that the proposed method is obviously superior both in vision and in PSNR.2. A new local adaptive threshold estimation method for image denoising based on the Nonsubsampled Contourlet Transform and the Principal Component Analysis is proposed in this paper. By utilizing noise energy based on PCA in NSCT domain,we propose a local adaptive threshold considering the clustering property of the coefficients for remote sensing image denoising. The simulation experimental results show that the proposed method can effectively reduce Gauss noise in remote sensing image and preserve image edg.