Real-Time Recognition of Road Traffic Sign U sing N ew Image Filter

This research is concemed with a new technology for the recognition of traffic sign from a motion picture using search region limits genetic algorithms (denoted simply by GA). Though former image pattern recognition has chiefly been studied only for an individual object， it is an advanced direction to recognize the object from a motion picture， along with the development of the visual system of the robot. However， it is generally difficult to recognize a specific object from scene image. Because there exist many other objects and unexpected objects expect a target in an image， and also， a degree of freedom for a relative position between a target and a video becomes large. In this paper， a new method is developed to recognize a traffic sign of a circle shape in a motion picture by application of GA. In order to recognize the specific object from a motion picture， it is necessary to separate the object from background. However， if the object is a traffic sign， it is difficult to separate with this sign from the scene image. Generally， in the detection of a traffic sign,there are many degrees of freedom for both size and background in the image. Then， a traffic sign must be explored in such difficult conditions. In this study， a new recognition method is proposed which detect a traffic sign by using a search region limits GA