An Improved Image Denoising Method Applied in Resisting Mixed Noise Based on MCA and Median Filter

Images are usually inevitably polluted by noise during the transmission and the reception of images. The reliability and the quality of an image can be reduced by the noise. Noise suppression is one of the important problems that can not be ignored in image processing. Through the past decades, various of denoising methods had been adopted and received good performance. These methods denoise a noisy image which is supposed to a Gaussian noisy image. But the common noise that exist in image are Gaussian noise and pepper and salt noise. Most of the denoising methods only considers Gaussian noise while ignoring pepper and salt noise. Denoise an image with both gaussian noise and pepper and salt noise via traditional methods can not remove noise completely. An improved image denoising method based on (Morphogonal Component Analysis) MCA and median filter is carried out in this paper to overcome above drawbacks. In this work, first we decompose the image into structure image, texture image and edge image by adopting improved MCA algorithm. Considering with the increase of the numbers of the iterations, pepper and salt noise mainly gathered in texture part. Denoising texture part via median filter can be viewed as a pre-process step. After pre-procession, then use corresponding methods to process each part. At last, add three denoised parts to become the final denoised image. Numerical results show our method can get better denoising performance both in PSNR value and visual effects.