An implementation of real-time hardware for moving object detection and discrimination

In this paper, we introduce real-time hardware implemented for the purpose of detecting the boundary points of a moving object and discriminating it from non-targets, which are basic functions of an unmanned surveillance system. The implemented hardware is composed of the two parts; one is a hard-wired logic part which computes spatio-temporal derivatives, and the other is a discrimination part including a micro-controller and memory devices. The object boundary is extracted by using a moving region and an edge map which can be detected by temporal derivative followed by morphological erosion and spatial derivative, respectively. We propose newly devised logic circuits for subtraction and thresholding, and spatio-temporal derivatives. Scarcely before individual hardware modules complete their functions, the second image capturing is done by performing the above functions in real-time.