Research on Salt and Pepper Noise Removal Method Based on Adaptive Fuzzy Median Filter

Aiming at the problem that the traditional median filtering algorithm can not well retain the image details while eliminating image noise, this paper proposes an adaptive fuzzy median filtering algorithm based on the mean of absolute average gradient difference. The whole filtering process is divided into three parts: coarse noise detection, fine noise detection and noise filtering. First, the extreme value method is used to perform rough noise detection on the image to be tested, which is divided into suspected noise points and signal points; then use this algorithm to construct the fuzzy membership function of the mean of absolute average gradient difference between the detected suspected noise points and the signal points in the window (including the filtered noise points), and use this function to perform secondary detection on the suspected noise points to determine Whether it is noise and fuzzy classification processing; finally, perform different filtering processing on the noise fuzzy classification results. Experimental results show that the algorithm proposed in this paper has significant improvements in PSNR, MSE, and SSIM compared to the traditional median filtering algorithm. Therefore, the algorithm in this paper can effectively remove the salt and pepper noise of the image while protecting the image texture details.