NSCT De-noising Algorithm with Image Partition and Non-local Means

摘要

In order to preserve the integrity of edge and detail information of image, a NSCT de-noising algorithm with image partition and Non-local Means is proposed. Since NSCT has the features of translation invariance and multi-directional selectivity and non-local means method can make full use of the redundant information, a combination of the two methods is conducted to get a remarkable de-noising effect. Firstly the image is divided into several blocks with the same size, then the non-local means method is introduced to smooth the image and preserve the detail information of each block, furthermore, the noise variance and de-noising threshold of each block will be calculated and NSCT is applied to de-noise each block, finally, the blocks are merged. The experimental results demonstrate that the new image de-noising algorithm achieves excellence performance.