Binocular Depth Perception of Computer-Generated Patterns

The question of how the two-dimensional projections of the visual world that are supplied to the left and right eyes are matched and combined to reveal the impression of depth is an extremely interesting one. Because of an incorrect analogy derived from measuring distances with 1125 1154 T H E BELT, SYSTEM TECHNICAL J O U R N A L , S E P T E M B E R 1D()0 a range finder, it is commonly thought that this problem is rather trivial. Admittedly, it is fairly simple to determine binocular parallax by aligning selected portions of an object in the left and right fields of a range finder and computing depth by trigonometrical calculations. The intriguing part of this problem is to explain the remarkable ability of humans to establish correspondence between complicated patterns in the two monocular fields. This pattern-matching process is the one being investigated here. It seems quite clear that patterns perceived in depth afford a promising means for exploring pattern-matching. However, it is well known that the perception of depth under familiar conditions is mediated by many complex cues, both binocular and monocular, which are not easily kept under the control of the experimenter. Thus, many previous explorations have used stereo pictures of familiar objects or line drawings, precluding the separation of interacting cues. The investigation reported here utilized patterns devoid of all cues except binocular parallax, by using artificially created stereo images with known topological properties.