Study And Implementation Of The Face Detection Algorithm For The Digital Monitor Images

These days, the digital monitoring systems that based on network are widely used in important sites for security's sake. In the system's working mode, the recorded data stream needs to be transmitted to the monitoring computer continuously through the image transmitting network. So the data stream is enormous and it becomes a great challenge for the data storage space and network load. If there is such a face detection algorithm, which can detect the face area in every frame of image and only the data of face area is saved, network load and the storage space needed for bank monitoring system will be greatly reduced.According to digital monitoring system's need, the algorithm used both skin and lip-color information and the faces are detected through the lip detection in candidated face areas. The author founded CAU-CS face database, by the statistics of experiments on samples of the database and the skin optimum threshold range accounting algorithm, the optimum threshold scope is defined. In the skin detection algorithm, the author compared the advantages and disadvantages of every color space and presented a method combing YCb'Cr' and YES color space. The above two points are the new ideas in this paper. And the face detection algorithm is also designed and implemented in this paper. The algorithm implements light compensating on monitoring images to remove the influence of the environment, such as lamp-house, color etc. To eliminate small hole in two values images, the algorithm implements close operation for improving the detection ratio.To prove its validity, the author tested the face detection algorithm using CAU-CS face database. The experiments showed that the algorithm is of real-time and has high detection ratio.