Title :
Saliency mapping in associative vision machine
Author :
Akerib, Avidan J. ; Ruhman, Smil
Author_Institution :
Dept. of Appl. Math. & Comput. Sci., Weizmann Inst. of Sci., Rehovot, Israel
Abstract :
A very effective measure of saliency, proposed by A. Sha´ashua and S. Ullman (1988), is based on curve length, continuity, and smoothness. Implemented as an iterative process to reduce complexity, it still takes 0.5 s/iteration on the Connection Machine. The authors present an associative algorithm more than three orders of magnitude faster when executed on the ARTVM (Associative Real Time Vision Machine). This is a classical associative architecture, adapted for vision and modular VLSI, that requires a very modest hardware complement (<100 chips)
Keywords :
computer vision; iterative methods; parallel architectures; parallel machines; real-time systems; ARTVM; Associative Real Time Vision Machine; associative vision machine; classical associative architecture; complexity; continuity; curve length; hardware complement; iterative process; modular VLSI; saliency mapping; smoothness; Attenuation; Computer science; Hardware; Image segmentation; Iterative algorithms; Length measurement; Mathematics; Shape measurement; Testing; Very large scale integration;
Conference_Titel :
Signals, Systems and Computers, 1992. 1992 Conference Record of The Twenty-Sixth Asilomar Conference on
Conference_Location :
Pacific Grove, CA
Print_ISBN :
0-8186-3160-0
DOI :
10.1109/ACSSC.1992.269092
