An Improved Median Filter Based on Automatic Parameter Tuning Approach

Image restoration is an important and widely studied problem in the area of computer vision and image processing. Various effective image-filtering strategies have been proposed, however a common drawback of some methods is that they usually require some a priori knowledge about the density of impulse noise corruption. A novel efficient improved median filter based on automatic parameter tuning approach is proposed in this paper for the restoration of images corrupted by impulse noise. The proposed method combines an improved median filtering algorithm that includes two parameters with the technique of automatic parameter tuning based on noise estimation. The parameters can be adaptively adjusted for minimizing the MSE value. As a key feature, the method did not require any a priori knowledge about the density of noise corruption. Comparative studies of the method and other methods for four different images are made and the advantages of the proposed method are fully demonstrated.