Title :
System-level design of specialized VLSI hardware for computing relative orientation
Author_Institution :
Artificial Intelligence Lab., MIT, Cambridge, MA, USA
fDate :
30 Nov-2 Dec 1992
Abstract :
Determining the relative orientation between the coordinate systems of two cameras is central to binocular stereo, as well as to the long range motion vision. The author proposes a system-level design for specialized hardware to compute relative orientation in real-time. The problem is difficult for two reasons: one is the nonlinearity of the equations which must be solved, and which may possess multiple solutions or local minima; the other is the difficulty of finding point correspondences. The major contribution of the paper is the theoretical analysis and simulation of simple methods, which map well to analog and digital hardware, for resolving both difficulties
Keywords :
VLSI; stereo image processing; VLSI hardware; binocular stereo vision; long range motion vision; relative orientation; system-level design; two cameras; Analog computers; Artificial intelligence; Circuits; Digital cameras; Hardware; Image edge detection; Nonlinear equations; Real time systems; System-level design; Very large scale integration;
Conference_Titel :
Applications of Computer Vision, Proceedings, 1992., IEEE Workshop on
Conference_Location :
Palm Springs, CA
Print_ISBN :
0-8186-2840-5
DOI :
10.1109/ACV.1992.240319