Identification of Salient Contours in Cluttered Images

We present a model of contour extraction in which the perceptual salience of contours arises from long-range interactions between orientation-selective filters. Ullman [19], Zucker [22, 23] and colleagues have previously shown that salient contours may be extracted from noisy images by using a number of heuristic features. Our algorithm is based on cortical mechanisms, and simulations show close agreement with results from recent anatomical, physiological and psychophysical studies including recent results of Field et al. [3], Kovács et al. [10, 11, 12], and Kapadia et al., [8]. The performance of the algorithm is demonstrated on a range of psychophysical stimuli and real images. 1. Introduction Contour extraction has been the focus of many previous computational studies, yet it remains a difficult problem in practice. As pointed out by Sha'ashua and Ullman [19], the relative salience of the contours in an image suggests something about the cortical mechanisms used to extract them. Sh..