Abstract: A novel filtering algorithm called adaptive directional weighted median (ADWM) filtering is proposed. The ideas of the directional filt ering and the weighted median filtering are combined with each other to construc t the ADWM filter model, which is a highly nonlinear image smoothing and enhance ment technique. The use of the variance of the moving window and the base varian ce supports the adaptivity of the ADWM filter. The ADWM filter has the features of directional filter, which is especially effective at preserving edges, and th e features of center weighted median (CWM) filter, which can reduce random noise , and moderate amounts of impulsive noise also, to some extent, as well as p reserve the details via adaptively determining the weight of the center pixe l according to the lowest variance of pixels within all the sub-windows.

Key words: adaptive directional weighted median filtering, direc tional filtering, filter, median filtering, weighted median filtering