The Role of Curvature in Shape Recognition
Curvature is a pervasive feature of the visual world, present in both natural and man-made objects. Evaluating the curvature of object boundaries and surface contours is therefore a fundamental component of visual processing. From an ecological perspective, analyses of natural scene statistics provided evidence for the prevalence of co-circular contours in our visual environment, which may provide important cues for the visual system. There is a long tradition of research investigating the role of curvature in shape and object perception. Evidence from psychophysics and neurophysiology suggests that the visual system contains mechanisms that are selectively sensitive to curvature and that these mechanisms may contribute to the formation of object representations. In this talk, I review behavioural and physiological evidence indicating that curvature plays an important role in shape perception, with particular emphasis on specific curvature points, such as local curvature extrema. I will focus primarily on psychophysical findings while relating them to current physiological knowledge and computational models of shape representation. The aim is to evaluate the extent to which curvature-based representations can account for shape perception and to discuss their limitations within a broader framework of object recognition.