This paper describes the use of configurable logic boards to carry out real-time stereo image matching. The main factors in achieving this high performance are the elimination of instruction and data fetching overheads associated with conventional processors and the use of a non-parametric matching measure. Many reported stereo-matching systems are based on numerically intensive measures such as normalized cross-correlation which is expensive in terms of logic. Non-parametric measures require only simple, parallelizable, functions such as comparators, counters and exclusive-or, and are thus better suited to FPGA implementation. Further advantage is obtained by being able to overlap processing with image input and output, and parallel computation of disparities.
CITATION STYLE
Dunn, P., & Corke, P. (1997). Real-time stereopsis using FPGAs. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 1304, pp. 400–409). Springer Verlag. https://doi.org/10.1007/3-540-63465-7_245
Mendeley helps you to discover research relevant for your work.