Purchase this article with an account.
Inna Tsirlin, Robert Allison, Laurie Wilcox; Is depth in monocular regions processed by disparity detectors? A computational analysis.. Journal of Vision 2012;12(9):215. doi: 10.1167/12.9.215.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Depth from binocular disparity relies on finding matching points in the images of the two eyes. However, not all points have a corresponding match since some regions are visible to one eye only. These regions, known as monocular occlusions, play an important role in stereoscopic depth perception supporting both qualitative and quantitative depth percepts. However, it is debated whether these percepts could be signaled by the activity of disparity detectors or require cells specifically tuned to detect monocular occlusions. The goal of the present work is to assess the degree to which model disparity detectors are able to compute the direction and the amount of depth perceived from monocular occlusions. It has been argued that disparity-selective neurons in V1 essentially perform a cross-correlation on the images of the two eyes. Consequently, we have applied a windowed cross-correlation algorithm to several monocular occlusion stimuli presented in the literature (see also Harris & Smith, VSS 2010). We computed depth maps and correlation profiles and measured the reliability and the strength of the disparity signal generated by cross-correlation. Our results show that although the algorithm is able to predict perceived depth in monocularly occluded regions for some stimuli, it fails to do so for others. Moreover, for virtually all monocularly occluded regions the reliability and the signal strength of depth estimates are low in comparison to estimates made in binocular regions. We also find that depth estimates for monocular areas are highly sensitive to the window size and the range of disparities used to compute the cross-correlation. We conclude that disparity detectors, at least those that perform cross-correlation, cannot account for all instances of depth perceived from monocular occlusions. A more complex mechanism, potentially involving monocular occlusion detectors, is required to account for depth in these stimuli.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only