Adaptative low-pass fuzzy for noise removal / Jorge A. Silva Centeno, Victor Haertel.

Satellite imagery often has random noise superimposed upon the brightness. Spatial low-pass filtering is commonly used to reduce such noise. This is a context dependent technique which takes into account the brightness of neighboring pixels, weighted according to their distance to the pixel filtered. This classical approach often results in image blurring. In order to circumvent this problem, a second weighing function is proposed in this paper. It implements the concept of fuzzy logic membership function, based on the brightness of each pixel within a neighborhood. The resulting combination of the two weighing procedures presented better results than usual weighing as a function of distance only. The proposed algorithm was implemented on a SUN SPARCSTATION 2 for testing purposes. Results of the procedure are presented..