A Method of Mixed Noise Suppression for Edge-preserving
Yang Dong-yun, Guohui Li · Harbin Ligong Daxue xuebao · 2005
Based on wavelet transform, a high frequency power preserving (HFPP) method is introduced for the edge detection of the images, and a mutual power cross (MPC) method is presented for the suppression of additive Gaussian white noise. In accordance with the edge detection and mixed noise suppression contradiction, a novel combined median filter and mutual power cross (MMPC) method is proposed. This new combined MMPC method is mainly based on that the class of median filters is suitable to the impulsive noise suppression and wavelet filters are suitable to additive Gaussian white noise suppression. The simulation results on Test images and Lena images show that the major edges of the images are detected well by the proposed HFPP method. The ability to suppress the mixed noise is improved by utilizing the combined MMPC method. The new method optimizes the tradeoff between the accuracy of edge detection and the ability of the noise suppression.