Two-dimensional recursive filtering algorithm with edge preserving properties and reduced numerical complexity

摘要

The filtering problem for noisy nonhomogeneous images is considered. A semicausal recursive filter with edge information is proposed. The filter is obtained as a combination of a one-dimensional (1-D) fixed-interval smoother and of a 1-D Kalman predictor, and involves a modest computational burden. Experimental results show the effectiveness of the new filtering scheme.