Computing the cost of occlusion

Recently, Cox et al. (1996, CVGIP: Image Understanding 63, 532-567) present ed a new dynamic programming-based stereo matching algorithm. The algorithm uses a parameter which represents the cost of occlusion. This cost is levi ed if the algorithm decides that two measurements, each from a different ca mera along corresponding epipolar lines, are not projections of the same po int in space. The occlusion cost is dependent on the standard deviation of the (Gaussian) sensor noise, sigma, and the probability of match detection, P-D. Under certain conditions such as low signal-to-noise ratio, the algor ithm of Cox et al. will declare occlusions where they do not exist. We offe r an alternative definition for the cost of occlusion, based on a decision- theoretic formulation for the matching process. This alternative improves t he performance of the matching algorithm. (C) 2000 Academic Press.