Total Variation Image De-noising Bases on the Improved Sobel Operator

Traditional Total Variation algorithms often ignore the images edge direction. In order to make up for the flaw of the algorithm existing , a novel denosing method based on direction Total Variation is introduced, on the basis of the combination of the gradient magnitude and orientation. Firstly, the pixels are divided into edge regions and non-edge regions by gradient magnitude. Secondly, the different 4-neighborhood pixels are found in different regions based on gradient orientation. Finally, various of the neighborhoods are analyzed by traditional Total Variation algorithms, so the proposed method can preserves image edge information during image de-noising. Experimental results show that the method combining edge direction information proposed in this paper can improve the neighborhood selection strategy in traditional Total Variation. It not only preserves edges and important details effectively, but also improves the PSNR and visual effect of