Abstract
Traditional stereo grouping models (eg Marr and Poggio, 1976 Science 15 283 - 287) have focused on the stereo correspondence problem the matching of the corresponding monocular inputs to obtain 3-D depth. Correct stereo correspondence is responsible for, eg, disparity capture (the
propagation of depth information from the boundaries to the centre of a depth plane to break, eg, the wallpaper illusion), and (depth) transparency. V2 cells were recently observed to exhibit disparity capture via contextual influences (Bakin et al, 2000
Journal of Neuroscience 20 8188 ^ 8198). Recent physiological data, however, revealed additional unexpected stereo grouping behaviour. Some V2 cells increase their responses to stimuli of their preferred depth when the stimuli within their receptive fields are at or near the boundary of a depth surface (von der Heydt et al, 2000
Vision Research 40 1955 ^ 1967). Such highlights to depth edges are seemingly
not required computationally merely to solve the correspondence problem. Computationally, these highlights make the boundaries of a depth surface more salient, serving pre-attentive segmentation and attracting visual attention. In special cases, they enable the psychophysically
observed perceptual pop-out of a target from a background of visually identical distractors at a different depth. To achieve the highlights, mutual inhibition between disparity-selective cells tuned to the same or similar depths is required. However, such mutual inhibition should impede the
computation for the correspondence problem, which requires mutual excitation, instead, between the same cells. In this work, I introduce the first computational model to address both stereo
correspondence and pre-attentive stereo segmentation. The computational mechanisms in the model are based on intracortical interactions in V2. I demonstrate that the model captures the following physiological and psychophysical phenomena: (i) depth edge highlighting, (ii) disparity capture, (iii) pop-out, and (iv) transparency.
