The patchwork engine: image segmentation from shape symmetries

We propose blind segmentation of images into shape-related 'patches' based on pre-calculated local symmetries (Van Tender, G.J. & Ejima, Y. (1999). (F orthcoming a) Flexible computation of shape symmetries. Submitted for publi cation) in shape boundary contours. First, lateral weights between all poin ts in the boundary contour map are assigned analogous to Euclidean distance maps in watershed segmentation (Beucher, S. & Lantejoul, C. (1979). Use of watersheds in contour detection. Proceedings of the International Workshop on Image Processing, CCETT, Rennes, France.). Lateral weights are then use d to: (1) extract local maxima in symmetries; (2) link maxima within locall y enclosed boundary contours; and (3) reconstruct shape contours using symm etry maxima as 'seeds'. The new model overcomes weaknesses of watershed seg mentation. The new model closes gaps in relatively more solid image contour s, but it is fundamentally different from methods based on contour interpol ation (Grossberg, S., Mingolla, E. & Todorove, D. (1989). A neural network architecture for preattentive vision, IEEE Transactions on Biomedical Engin eering 36, 65-84; Heitger, F. & von der Heydt, R. (1993). A computational m odel of neural contour processing: figure-ground segregation and illusory c ontours. Proceedings of the Fourth International Conference on Computer Vis ion, IEEE Computer Society Press, Washington D.C. (pp. 32-40)). Images are segmented into shape-relevant color-by-number-like patches which compare we ll to related methods (Gauch, J. & Pizer, M. (1993). The intensity axis of symmetry and its application to image segmentation, IEEE Transactions on Pa ttern Analysis and Machine Intelligence, 15 (8), 753-770; Ilg, W. & Ogniewi cz, R. (1995). The application of Voronoi skeletons to perceptual grouping in line images, Proceedings of the 11th International Conference on Pattern Recognition, The Hague, The Netherlands, pp. 382-385; Zhu, S.C. & Yuille, A.L. (1996) FORMS: a flexible object recognition and modeling system, Inter national Journal of Computer Vision, 20 (3), 187-212.). Two primitive opera tions, comparison and merging of patches, are proposed as drives for exposi ng more global shape contours from patches. We conclude that symmetry goes beyond abstract shape morphology: it can contribute to figure-ground segmen tation in early vision and form part of primitive operations needed to crea te hypotheses of complex shape. (C) 2000 Elsevier Science Ltd. All rights r eserved.