Based on the workings of visual cortical area V1, a model for an architecture for early computer vision is proposed. We propose to do image processing for computer vision on the basis of a combined map, of edge orientation and ocular-dominance, rather than on the basis of an edge map alone. In particular, the combined representation of edge orientation and ocular dominance is proposed for the computation of stereoscopic disparity.
Research on the role of human stereopsis has largely focused on laboratory studies that control or eliminate other cues to depth. However, in everyday environments we rarely rely on a single source of depth information. Despite this, few studies have assessed the impact of binocular vision on depth judgements in real-world scenarios presented in simulation. Here we conducted a series of experiments to determine if, and to what extent, stereoscopic depth provides a benefit for tasks commonly performed by helicopter aircrew. We assessed the impact of binocular vision and stereopsis on perception of (1) relative and (2) absolute distance above the ground (altitude) using natural and simulated stereoscopic-3D (S3D) imagery. The results showed that, consistent with the literature, binocular vision provides very weak input to absolute altitude estimates at high altitudes (10-100ft). In contrast, estimates of relative altitude at low altitudes (0-5ft) were critically dependent on stereopsis, irrespective of terrain type. These findings are consistent with the view that stereopsis provides important information for altitude judgments when close to the ground; while at high altitudes these judgments are based primarily on the perception of 2D cues.