Multi-scale edge finding in noisy pictures

Finding edges in digital images is a very basic and important task. This paper describes a new technique to address this problem. The proposed technique tries to solve two of the main difficulties that edge finding algorithms must face: uncertainty due to noise artifacts and combination of details detected at different scales. In particular we adapt a fuzzy arithmetic approach called “Fuzzy Marching Squares” (G.Gallo, S.Spinello, Thresholding and fast iso-contour extraction with fuzzy arithmetic, Pattern Recognition Letters 21 (2000) 31-44) to obtain the edge map of a digital picture. Our map assembles together information about uncertainty and scale. Comparisons with results obtained with Canny’s algorithm show that our approach is particularly suitable in presence of additive noise. As for complexity the proposed approach is comparable with the application of traditional median filter.