MIXED-SIGNAL VLSI ARCHITECTURE FOR REAL-TIME COMPUTER VISION

This paper presents the architecture of a computer vision system targe ted for real-time robot vision and pattern recognition applications, T he proposed mixed-signal very large scale integration (VLSI) architect ure integrates photo-transduction with low-and medium-level processing such as multi-resolution edge extraction, scale-space integration, ed ge tracking, dominant point extraction, and database generation. Its h igh performance stems from a custom CMOS smart image sensor providing parallel access to illuminance data and a set of parallel analog filte rs performing multi-resolution edge extraction, We have also developed a digital controller which manages data flow between the processing m odules of the system and which constructs a database of the observed s cene under the supervision of a digital signal processor (DSP) unit, T his database describes relevant object contours as a linked list of li near segments and circular arcs with precomputed local and global prop erties. Such a token description of the scene is suitable for robot vi sion and pattern recognition applications, since it significantly comp resses the amount of data to be processed by further high-level algori thms, Experimental results obtained with the current prototype of the system are very promising, with the complete process, from image acqui sition to scene database creation, performed in less than a second. (C ) 1997 Academic Press Limited.