Rapid Prototyping Platform for Reconfigurable Image Processing

Image processing and computer vision is being used in a variety of applications including applications in embedded devices (surveillance, quality control, traffic monitoring, etc.). In these applications, the data are used in real-time (in the traffic monitoring and decision-making sensors for example). Real-time video processing is computationally demanding, making microprocessor-based processing impossible. Moreover, microprocessor DSPs is energy inefficient, which can be a problem in industrial power-limited embedded systems. The trend of exploiting embedded low-power and low-cost devices for image processing and computer vision is becoming stronger. Using reconfigurable computing, it is possible to achieve high computational performance while the target system remains relatively inexpensive. This paper outlines a proposed architecture for an integrated partial rapid prototyping dynamic reconfigurable system-on-a-chip, targeted at embedded real-time video and image processing.