Application of VLSI for image processing

Real time signal processing implemented with VLSI chips is used to find an aimpoint on one object in a set of objects that were imaged in the focal plane. The authors explain the function of each of the algorithm steps needed and show that a VLSI implementation using either a SIMD (single-instruction-multiple-data) or an RPA (reconfigurable processor array) can be used to process a 128*128 image at a 10 ms cycle rate with an effective throughput of 100 MOPS. The alternative RPA solution, which is based on existing VLSI LINC (link and interconnect chip) and ALU SlP (scan line processor) chips, is presented. The RPA approach can only achieve the cycle time constraint when the clustering is performed by a specialized VLSI chip. Although the image processing function is more compatible with the SIMD approach, the RPA approach using the clustering chip is shown to require far fewer VLSI chips.<>