Computer vision algorithms on reconfigurable logic arrays

Computer vision algorithms are natural candidates for high performance comp uting systems. Algorithms in computer vision are characterized by complex a nd repetitive operations on large amounts of data involving a variety of da ta interactions (e.g., point operations, neighborhood operations, global op erations). In this paper, we describe the use of the custom computing appro ach to meet the computation and communication needs of computer vision algo rithms. By customizing hardware architecture at the instruction level for e very application, the optimal grain size needed for the problem at hand and the instruction granularity can be matched. A custom computing approach ca n also reuse the same hardware by reconfiguring at the software level for d ifferent levels of the computer vision application. We demonstrate the adva ntages of our approach using Splash 2-a Xilinx 4010-based custom computer.