Applied Research On Image Recognition Technology In Intelligent Video Surveillance System

Intelligence of the video surveillance system is mainly reflected in the intelligent recognition function.The function of image recognition is an important symbol of the development of the video surveillance system. The core of intelligent video surveillance technology is the image recognition algorithm. As face detection and recognition technology are typical and practical, this thesis choose face image as the recognition target.In this thesis, the Client/Server mode was used to design the video surveillance system. On the client development board the image acquisition and video transmission tasks were completed and on the server computer the algorithm was realized based on OpenCV, thereby face detection and recognition was realized in video surveillance system.More specifically, for the first, video frames was captured based on DaVinci technology and transmitted to the computer on RTP/RTCP protocol; for the second, the Harr-like features of faces were extracted form the video, the integral image of the feature was calculated and the adaboost cascade classifier was designed to detect the face target, experiments shows that the detection rate can meet face recognition demand; at last, the statistic features of faces were extracted, principal component analysis method was used to reduce dimensions of the face features and support vector machines was used to design the classifier and recognize the faces, experiments shows that algorithm is easily to realize and the recognition rate is better than90%.In conclusion, this thesis discussed the whole architecture, core algorithm and key technology when image recognition algorithm was realized in the video surveillance system. From the process that the intelligent algorithm was preliminary applied we can see that this thesis realize face detection and recognition in video surveillance system successfully. Our methods are practical and universal.