Research Of Face Recognition Technology Based On Video Surveillance Images

The department of Social Order, Public Security, Criminal Investigation, Law Enforcement Agencies and SecurityCompanies and other units are facing increasingly heavy video surveillance alone is artificial to monitor tens of thousands of camera and monitor screen to complete the search and identification of the monitored objects such as the task has become increasingly difficult.In the video surveillance system, how many camera and imaging machine, the query target person information needed by the security personnel.Clearly, in accordance with traditional practice, but to take the "human wave tactics" to invest substantial human and material resources to view all the video data. Find the help of manpower to bring the following drawbacks:(1) Slow:the human energy is limited, not round the sleepless go see a large number of video surveillance.(2) Cost:today’s large-scale monitoring system, the distribution of the with the thousands of cameras, insufficient staffing artificial visual way to monitor, will be a valuable cost.(3) Delay in the timing:it is easy because of the failure to grasp the trends of the target person, missed a good opportunity to stop or solve the case, and to avoid the major events that may occur.This article is based on this starting point on the surveillance video of the movement target partition and extract the movement target region, Face Detection and extraction of facial feature, complete the facial feature comparison, provide a target person appeared the event, the camera’s The surveillance video of the location containing the target person, and technology integration form a complete solution for intelligent video surveillance, Security Personnel play an important supporting role.This article is based on the monitoring requirements of the video face recognition technology to build a surveillance video to find the face recognition system after the segment monitoring based on temporal differencing algorithm in video sequences of moving targets and extract the region of the moving target, and then through Adaboost Face Detection Algorithm to complete the movement of the target area of face detection, feature extraction and classification.Then, using a face detection based on improved Adaboost Face Detection Algorithm on these campaigns target area, feature extraction and classification and identification operation. analysis, and tests showed that the above algorithm applies to monitor the video face detection prototype system based on improved interframe difference and Adaboost Algorithm combining.Then, building and afterwards finding the face recognition system based on the surveillance video, and details of elements of system architecture and system, and relevant components of the surveillance video of a prototype face recognition system was introduced.Finally, building the prototype system environment, and this monitor video face detection prototype system functionality, performance and stability testing. The test results show that this prototype system is fast, simple, non-intrusive and does not require people actively cooperate with the characteristics, with the cost, economic, and scalability of good features. The innovation of this paper is as follows:1.Interframe difference method and the Adaboost algorithm combined and used to monitor the video processing, the introduction of a find the face recognition system based on the surveillance video after the scene is about to concern tune a good camera to record and save, will The image of the target person into the prototype system to start than if the relevant records, the system will bring up the video clip that appears in the target person.2.For the identification system in the surveillance video after Find improved interframe difference and Adaboost algorithm to adapt to real-time requirements, so a short period of time to complete the targeted person is to find, and the target person’s time, the position of the camera, as well as video playback, and other information of the target person.