A Discontinuity Adaptive Prior for Image Denoising

Abstract

The presence of noise in digital images degrades the visual quality by corrupting the information associated with the image. The aim of denoising is to restore an image from its noisy version by preserving signal information. In this paper, we are considering an image corrupted by additive Gaussian noise. The image is modeled as Markov random field (MRF) and an estimation of maximum-a-posteriori (MAP) is obtained using graduated non-convexity. The results are compared with other spatial domain filtering methods. The discontinuity adaptive prior helps in preserving edge information. The results suggest that proposed method has an improved performance.