Pixels classification in noisy digital pictures using fuzzy arithmetic

This paper presents a new technique to extract, in noisy digital pictures, regions whose pixels fall, with a degree of uncetainty, in a given range of gray levels. The proposed method uses fuzzy numbers to describe in a compact way, at the early vision stage, the relevant information of the picture together with the uncertainty due to noise. This fuzzy model of the original picture is hence interrogated with a Marching-Cube-like algorithm to obtain, for a specified level of presumption, the pixels in a prescribed range. The quality of the obtained results is comparable with those obtained with more traditional, but less efficient, non-linear smoothing techniques