Blind Color Image Denoising Using Low Rank Matrix Minimization

摘要

This paper presents a simple approach for reducing the Additive White Gaussian Noise (AWGN) on noisy color image. The proposed approach exploits the effectiveness of Low Rank Matrix Minimization (LRMM) technique for suppressing the occurred noise. However, the LRMM has a slight limitation on the image denoising process. It requires the prior knowledge about the noise level. To overcome this aforementioned problem, we develop a simple noise level estimation for color image using the Singular Value Decomposition (SVD). Our proposed method performs blind image denoising task over LRMM setting and SVD-based estimated noise level. As documented in the experimental section, the proposed method yields a promising result for the blind image denoising task.