Surface segmentation based on perceptual grouping

Conference Paper · February 1999with3 Reads
DOI: 10.1109/ICIAP.1999.797616 · Source: IEEE Xplore
Conference: Image Analysis and Processing, 1999. Proceedings. International Conference on

    Abstract

    This paper deals with the achievement of the best contour grouping
    of an image from the point of view of the probability that the grouping
    corresponds to either an object or an object surface from the scene. For
    the grouping of the contours, the principle of non-accidentally has been
    used. This principle rests on psychological mechanisms of perceptual
    organization such as proximity, co-linearity, co-curvilinearity,
    convexity and closure. Neither do all mechanisms of perceptual
    organization have the same degree of importance, nor are the primitives
    they process equal. In consequence, we propose a novel approach for
    perceptual grouping that takes into account the sequential order in the
    application of the different mechanisms for grouping of primitives, so
    that it will change in function the relative importance of the
    perceptual mechanisms in each particular case