Extracting long contour by using the competitive layer model of the Lotka-Volterra recurrent neural networks

Abstract

It is well known that contours in image are salient when a series of edge elements are aligned in a collinear or co-circular fashion. In this paper, a contour extractor is constructed to extract contours by using the competitive layer mode implemented by Lotka-Volterra recurrent neural networks. This extractor can bind all edge elements which belong to a contour onto a layer. In order to extract long contours, by moving the established contour extractor along a long contour, then the long contour can be extracted segment by segment. Experiments show that the proposed method can extract long contour properly from images.