Image adaptive filtering based on the improved Alpha-trimmed mean algorithm

Wangcheng Cao, Lin Zheng, Kuiling Dong 

School of Engineering, Mudanjiang Normal University, Mudanjiang 157011, Heilongjiang, China

Filtering is an important research direction of image processing. In view of the characteristics of image noise, an adaptive image filtering algorithm is proposed based on the improved Alpha-trimmed mean algorithm. The algorithm dynamically selects parameter d of Alpha-trimmed mean algorithm through the calculation of the pixel correlation within the filter window, so that the algorithm can filter accordingly with the degree of noise. The experiment results show that the filtering effect of the proposed algorithm is excellent for the images corrupted by Gaussian noise, salt and pepper noise or mixed noise, and this algorithm is also capable of maintaining the detailed information of the original image.