Abstract
These days, the demand of security video devices such as surveillance camera, video recorder and so on has quickly increased in accordance with the increasing number of crimes. The human face recognition technology in images is almost established. However, if the face is covered with sunglasses and/or a mask, face features are loss and the current face detection technology cannot be applied. Therefore, we have made a database to detect the suspicious people who put on sunglasses and/or a mask, and have established the detection method for the suspicious people by machine learning based on Haar-like features. Here, the only detection using the device based on Haar-like features causes false detections, so that we have improved detection precision by narrowing based on brightness features. In addition, we have compared the discrimination ability by using several detection devices made from some database.
