The Application Of Face Recognition In Apartments Video Surveillance

Video monitoring is an important part of safe guard system, widely used in government departments, finance, public security, transportation, electricity and other departments and industries,but most of the video surveillance only used for record,rarely people process the image in the video surveillance,therefore, the prevalence of intelligent applications is not high.Face recognition technology, as a highly secure biometrics technology, primarily depend on computers processing video images for identification and determination,to achieve authentication.This thesis aims at identifing people information in student apartments for video surveillance, by applying the basic method of face detection and face recognization,matching students will appear in the video image with the samples in self image face database, to identify the function of student information.To attain a better detection result for face detection,the method in this article need prior processing before detection. Skin color segmentation method to detect is faster but not high,while Adaboost detection algorithm is slower but high. Combine the advantages of these two methods to detect human faces and apply to daily life can achieve better and higher detection rate.The mainly applied method in face recognition process is principal component analysis(PCA)method. The distinguish feature of this method is the use of low-dimensional feature vectors to show the original sample information. Researchers need to project the images need to be recognized to feature vector space,and to realize face recognition according to suitability,it has better recognition accuracy.On Eclipse development platform,using cross-platform computer vision library OpenCV,combined with face detection and recognition methods in this article,students apartments being the objectives,have designed a friendly and simple user interface to implement identification system. The test results shows that the system has successfully implement identity authentication of people in the video surveillance,but the recognition accuracy of the system requires continuous improvement and perfection in real life application.