Removal of Gaussian noise with non-uniform total variation partition

Abstract

Non-uniform Partition (NUP) and Total Variation (TV) are both effective Gaussian denoising algorithms than many other common denoising algorithms, but some defects are still existing in them thus reducing their denoising performance. In this paper, a novel Gaussian denoising algorithm based on NUP and TV is proposed. TV is a global denoising algorithm which uses the same parameters and criteria to remove noise no matter it is a smooth or complex region, so its texture detail recovering ability is not strong. Square effect is the main problem in NUP denoising. In this proposed algorithm, TV and NUP are combined together so that TV denoising can be applied to the partitions with different complexity. And by applying the Interpolation Relaxation technique, the square effect in NUP is improved. The experimental results show that the proposed algorithm owns better denoising performance than the original NUP and TV denoising algorithm.