An Improved WNNM Algorithm for Image Denoising

Abstract

The traditional weighted nuclear norm minimization (WNNM) has excellent performance for the removal of non-sparse noise such as Gaussian noise, but attains bad performance for the removing of salt&pepper noise and mixed noise of Gaussian noise and salt&pepper noise. This paper proposes an improved WNNM image denoising algorithm. WNNM can effectively remove non-sparse noise such as Gaussian noise and adaptive median filtering algorithm can effectively remove sparse noise such as salt and pepper noise; the improved algorithm combines the characteristics of WNNM and adaptive median filtering. The experimental data demonstrate that the improved WNNM algorithm has better denoising effect than WNNM algorithm.