Person Recognition by Stereo Model-based Matching

This paper describes a method for detecting persons and measuring the 3D position in real time by stereo camera. We employ an unprocessed input color image termed as raw-image to execute recognition process by model-based matching method. A person on the image is detected by model-based matching and genetic algorithm. Face is defined as a circular skin color model for times of the image is evaluated by the face model. The maximum of evaluation value is searched by genetic algorithm. The position with the maximum evaluation value is detected as target person's face. The distance from the camera to the person is calculated by the disparity of detected person on the stereo images. The method obtains the person's position and height. Experimental results show the effectiveness of our proposed system