Adaptive Iterative Truncated Arithmetic Mean Filter in Image Denoising

Aimed at the excellence and shortcoming in noise attenuation and edge preservation of the arithmetic mean and the order statistical median, a new iterative algorithm that truncates the extreme values of samples in the filter window to a dynamic threshold combined with an adaptive selection of the filter window size based on noise detection is proposed in this paper. Stopping the iteration early, the proposed filter owns merits of both the mean and median filters in coping with "-contaminated Gaussian noise. The superiority and flexibility of the proposed Adaptive iterative truncated mean (Adaptive-ITM) filters are experimentally verified on real images corrupted by epsilon-contaminated Gaussian and alpha-stable noise.