Comparison of color median filters
Philippe Colantoni, Alain Trémeau, Bernard Laget · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1998
This paper compares different median filter methods to perform noise filtering and color image enhancement. The median filter performs a nonlinear filtering operation according to a window which moves over a signal, and computes at each point, the median value of the data within the window. This value is then taken as the output. In this paper, we have used a 3 dimensional signal: a RGB color. To find this median value, we need to define an order relation between all the points of the color data set. We propose to test 3 methods: one using the global principal component transform of the color image, one using the local principal component of the current window, one using the bit mixing, which transforms the 3D signal (the RGB components) directly in one 1D signal by a bit mixing.