(1.School of Information and Artificial Intelligence, Nanchang Institute of Science and Technology, Nanchang 330108, China;2.School of Materials Science and Engineering, Wuhan University of Technology, Wuhan 430205, China)
Abstract:Aiming for more effective image restoration from impulse noise, an adaptive and iterative deviation filter is proposed, which uses a noise detector employing the intensity feature of impulse noise as well as local deviation and local statistics for noise detection, and an adaptive and iterative technique for noise removal, which removes noise by referring to the local deviation of neighboring pixels, taking one neighboring pixel with the smallest deviation with neighboring noise free pixels as the new intensity of noisy pixel. The noise removal technique performs adaptively and iteratively, so as to cater for various noise densities and local inhomogeneity of noise distribution, and make the most of the previously processed results. In terms of peak signal to noise ratio, edge preservation index, visual representation, and computational time, experimental results reveal and verify that the proposed method has better capability in noise removal and image details preservation than the state-of-the-art filters.