Reduced complexity algorithm for approximate vector medians of three-dimensional discrete sets
Junwu Kan · Ha'erbin gongye daxue xuebao · 2009
A fast running algorithm based on the local window histogram was employed to implement approximate vector median filtering for multichannel image processing. The computational complexity of the algorithm is only O(n),so the median computation is not a bottleneck anymore for large image filters. Moreover,we prove that if the three-dimensional discrete set satisfies some connection conditions,the median obtained is a vector median. Otherwise,it is a scalar median of the window. The experimental results show that by means of the PSNR,SNR,MAE,MSE,NCD and NMSE coefficients,the effectiveness of the algorithm is almost as the same as VMF for the reduction of mutichannel impulsive noise in color images.