Abnormal Face Location And Recognition

Abnormal human face location and recognition is brought forward by the automatic surveillance and alarm system that is applied in the halls of banks or import indoor places. It aims at endowing the video surveillance system with intelligence. The video surveillance system, as a visible surveillance approach, is widely employed in the public security field. The system continuously captures the images of the monitored scene by a camera, perceives the invasion of irregular people with a mask on its face, a respirator on its mouth or a cap on its head and so on by the detection and recognition of the computer as early as possible and automatically alarms the remote control center according to the setting of the system.The dissertation proposed some methods to distinguish normal and abnormal human faces. After human head located, to distinguish the abnormal and normal faces, many characteristics between the normal and abnormal human face are investigated. In preprocessing, it firstly normalized face images to same brightness and contrast and secondly located the position of eyebrow and centerline of the face. After preprocessing, some new and practical methods and relative algorithms were proposed to recognize whether the sunglasses was in the eye region by analyzing the shape of center line of face and by analyzing statistic characteristics of gray values, whether respirator was in the face by detecting the horizontal and straight lines, and especially one method based PCA (Principal Component Analysis) was researched finally.