Cite as:
Bertamini M, Lawson R, 2005, "Segmentation based on binocular disparity and convexity" Perception 34 ECVP Abstract Supplement

Segmentation based on binocular disparity and convexity

M Bertamini, R Lawson

Figure - ground segmentation enables contour curvature to be classified as either convex or concave, and it is known that convexity and concavity information affects human performance. At the same time, it is believed that in the process of segmentation itself there is a bias to maximise convexity [Kanizsa and Gerbino, 1976, in Vision and Artifact Ed. M Henle (New York: Springer) pp 25 - 32]. We report a series of experiments using random-dot stereograms (RDSs) in which observers discriminate which of two areas is nearer in depth. This leads to one half of the display to be seen as figure and the other as ground. Binocular disparity in the RDSs defines a correct response. In addition, by using an aperture, only one segment of the contour that separates the two areas is visible. The task is therefore a local judgment. By contrast, previous work in the literature used ambiguous stimuli with no single, correct interpretation, and complex shapes bounded by contours with a mix of local convexities and concavities. We hypothesised that if the system has a bias to assign figural status to the convex side of a local contour, performance will be better when binocular disparity is consistent with this bias, ie when the convex area is nearer in depth. Results confirm this hypothesis.

These web-based abstracts are provided for ease of seaching and access, but certain aspects (such as as mathematics) may not appear in their optimum form. For the final published version of this abstract, please see
ECVP 2005 Abstract Supplement (complete) size: 2356 Kb