Massively parallel image processing system for intelligent transportation system applications

This paper presents a real-time image processing system for use in intelligent transportation systems. The system, run from a desktop computer, incorporates high-density pixel parallel (HDPP) processor chips into a programmable framework. By utilizing the system in a time-sharing fashion, multiple low-level image processing tasks can be implemented on the HDPP hardware. Examples of applications that have been implemented on the system and executed in real time include optical flow, template matching, and stereo vision.