Abstract: Digital images may be disturbed by different noises during acquisition and transfer processes. Treatment methods are varied with the characteristics of noises. It is interesting to develop algorithms to recover the images from strong noises, because the traditional filtering algorithms do not work well in strong noises. A new filtering algorithm combined median filtering with local area pixel clustering is proposed to filter strong Cauchy noises from digital images. Based on the processing of three images with Cauchy noise, it is demonstrated that the algorithm can obtain more details of images than the median filtering. In comparison with other methods, our filtering algorithm is an effective strong noise removal filtering algorithm because it protects image details and the edges significantly.

Key words: Image denoising, Strong noise, Median filtering, K-means clustering