Removal of fixed valued impulse noise using global noise statistics based adaptive histogram fuzzy filter

In this paper, a filtering technique based on the global noise statistics has been proposed to remove fixed valued impulse noise from color images. In this proposed filter, the size of the fuzzification window adapts itself according to the global noise density of the corrupted images. This global noise density is estimated using the statistics based on the deviation of the maximum and minimum intensity values. The filter takes into consideration the correlation between different color channels and operates only on noisy pixels. Moreover, vector median filter has been used at mid and high density noise for proper restoration. Experimental results show that the proposed algorithm is able to restore corrupted image in a more effective manner when compared to most of the existing filters with moderate computational complexity.