Fast image filtering via adaptive noise detection

This work develops a spatially white Gaussian noise detection algorithm for blind image filtering. First, the noise component as well as the noise level are estimated by using the local statistics from an observed degraded image. Then, the first order Markov Random Field is effectively used to control the noise detection performance. Finally, pixels, the adaptive weighted filter with the resizable patches is adopted to restore the detected noisy pixels. Numerous simulations have been conducted to demonstrate the effectiveness of the proposed method.