An Algorithm for Remote Sensing Image Denoising Based on the Combination of the Improved BiShrink and DTCWT

摘要

By considering the strong correlation between wavelet coefficients of the actual image, while bivariate model is only a statistical model for the interscale dependency of wavelet coefficient with parent coefficient, without taking into account the correlation of adjacent coefficient. Therefore, based on the shift-invariance and better directionality of the dual-tree complex wavelet transfer (DTCWT), and incorporating neighboring wavelet coefficients with BiShrink, a novel BiShrink threshold and DTCWT remote sensing image denoising method is presented.Experimental results show the proposed algorithm gets better PSNR than other methods mentioned, observably.In terms of visual quality, the proposed algorithm can get the images with more details, smooth profiles and aliasing is restricted.