Robust and optimal use of information in stereo vision

Differences between the left and right eye's views of the world carry infor mation about three-dimensional scene structure and about the position of th e eyes in the head The contemporary Bayesian approach to perception(1,2) im plies that human performance in using this source of eye-position informati on can be analysed most usefully by comparison with the performance of a st atistically optimal observer. Here we argue that the comparison observer sh ould also be statistically robust, and we find that this requirement leads to qualitatively new behaviours. For example, when presented with a class o f stereoscopic stimuli containing inconsistent information about eccentrici ty of gaze, estimates of this gaze parameter recorded from one robust ideal observer bifurcate at a critical value of stimulus inconsistency. We repor t an experiment in which human observers also show this phenomenon and we u se the experimentally determined critical value to estimate the vertical ac uity of the visual system. The Bayesian analysis also provides a highly rel iable and biologically plausible algorithm that can recover eye positions e ven before the classic stereo-correspondence problem is solved, that is, be fore deciding which features in the left and right images are to be matched .