Real-time image video edge enhancement using hardware convolver

Abstract

The system described is a real-time edge detector for use in image processing. The edge-detection algorithm is a 3/spl times/3 Sobel operator and adopts a new circuit of adding to orientation information. The hardware is designed to work at the 10 MHz pixel rate, and will accommodate images of 512/spl times/512 pixels at frames/s with noninterlaced video, or of up to 512/spl times/290 pixels with interlaced scanning. The hardware is pipelined and paralleled to achieve the required speed. 20 ms is needed to process a 512/spl times/290 pixels grey-level image. It may be used in military reconnaissance, TV real-time tracking or industry inspection.<>