탑승장 자세 인식을 위한 머리 검출 알고리듬

For a vision-based occupant s pose recognition system, this paper presents a new algorithm that can detect the head of the car occupant. Head detection is necessary for occupant's pose recognition in the car, since the position of occupant's head provides valuable information, such as his pose, size, position, and so on. We use the head-shoulder models and support vector machines. Given the variable illumination conditions within the car, the color information for detecting the head (or face) is not sufficient. Our method is based upon using only the grey image, since the infrared illumination could be utilized in the night. Although it is known that SVM could be useful for detection the face, such method can be slower when the size of the training set images is increased to cover diverse pose and size variation of the face. Since the contours in our study are lighter than the conventional face images in terms of SVM processing, it fits to the embedded system installed in the car. Results suggest that the head-shoulder contour model and support vector machines for detection the occupant's head could find a few useful applications.