An architecture for an integrated vision system is discussed. It is shown how various steps involved in an integrated vision system which consist of low-level, high-level, and hybrid algorithms can be efficiently mapped in an integrated environment. In particular, a stereo vision algorithm to extract 3-D object description from a set of 2-D images is considered. The emphasis is on using a small number of powerful processors concentrated in clusters and connected via a flexible, reconfigurable, and programmable crossbar. The issues considered are mapping algorithms independent of problem size, minimizing communication, efficient pipelining of tasks, and load balancing to evenly distribute the computation. It is shown why the architecture is efficient as an integrated vision system. Furthermore, it is shown how various steps of the algorithm can be mapped onto the architecture.
|Original language||English (US)|
|Number of pages||5|
|Journal||Proceedings of the International Conference on Parallel Processing|
|State||Published - Dec 1 1988|
ASJC Scopus subject areas
- Hardware and Architecture