An Adaptive Filtering Algorithm for Salt and Pepper Noise Based on Local Statistic Information

摘要

In order to filter out the salt and pepper noise in digital images,an adaptive algorithm is proposed based on switching filter.The algorithm includes three steps:noise detection,noise filter and error detection pixel correction.Firstly,noise was detected by the improved neighborhood of extreme value method,and noise marking matrix was established.Then,the noise was filtered out by gradient filter in the light polluted areas,by median filter in the moderately polluted areas,and by weighted mean filter in heavily polluted areas.Finally,the filter image was compared with noise image to correct mistaken filtered signal point.The results show the method has good adaptation in filtering variant degrees of noise polluted images.The filter image is better than the traditional median filter and some improved algorithm in terms of visual effects,or the effects of data recovery.