Image denoising using K-SVD and non-local means

This paper proposes an image denoising method, which exploit the non-local mean (NLM) algorithm and the sparse representation of images. The sparseness is computed by K-SVD and combined with the non-local mean algorithm. Images (Lena, House, Peppers, and Barbaba) with various noise levels (sigma =10, 20, 30, 40, and 50) are used to test the proposed method. The experimental results show that the NLM algorithm only performs better at the low noise level, while the proposed method performs better within a large range noise levels. The PSNR´s means of total images and all noises are 27.1712 and 27.7262 for the NLM and the proposed method. PSNR of the proposed method is 2% more than that of NLM algorithm. This indicates the proposed method performs better.