An Efficient Denoising Approach for Random-Valued Impulse Noise in Digital Images

Abstract

An efficient edge-preserving algorithm for removal of random-valued impulse noise from corrupted images is proposed in this paper. We employ an efficient impulse noise detector to detect the noisy pixels, and an edge-preserving filter to reconstruct the intensity values of noisy pixels. Extensive experimental results show that the proposed technique not only preserves the edge features, but also obtains excellent performances in terms of quantitative evaluation and visual quality. Since our algorithm requires only low computational complexity, it is very suitable to be applied to many real-time applications.