Research On AR Panoramic Linkage Monitoring System

Video surveillance system has intuitive and accurate content,strong timeliness,and can effectively prevent and crack down on criminal incidents as well as prevent various disasters and accidents.As an important part of comprehensive management of social security,video surveillance system has been widely used in various scenes.However,the traditional video surveillance system has monotonous content,lack of contact between multiple systems and mutual separation,due to the existence of many problems,it has been unable to meet the needs of today.As a hot field of computer vision,augmented reality technology has developed rapidly in recent years.With the maturity of technology,it has been widely used in various fields of society.In this paper,augmented reality technology is combined with video surveillance system,and the design scheme of AR panoramic linkage monitoring system is proposed,and the key technologies of the system are studied.The main research work is as follows:(1)The realization principle of augmented reality 3D registration technology is studied.Aiming at the problems that marker-based 3D registration method is easy to be affected by illumination and has poor anti-occlusion ability,the marker-less 3D registration method is studied.By constructing the nonlinear scale space,and then extracting the image features,the unmarked 3D registration method is realized by identifying the target through feature matching.Experiments are designed to compare the comprehensive performance of several mainstream feature extraction algorithms applied in 3D registration.(2)In order to solve the problem of 3D registration failure caused by image feature matching failure in large-scale difference monitoring image matching,the principle of scale invariant feature,namely image scale space theory,is studied in this paper,and a feature matching method of large-scale difference image based on vocabulary tree is studied.Firstly,the vocabulary tree algorithm is used to compact code the image scale space,and then the matching responses among all the scale levels in the scale space are calculated to obtain the scaling scale ratio among the images to be matched.The scaling scale ratio is used to limit the search space during feature matching,and finally the correct feature correspondence relationship is obtained.The algorithm improves the image feature matching performance in the case of large scale difference.(3)In order to solve the problem of 3D registration failure caused by image feature matching failure in large view point difference monitoring image matching.In this paper,the principle of the affine invariant features are studied,and fully affine invariant algorithm framework of ASIFT algorithm are summarized.The framework of the completely affine invariant algorithm is applied to the above algorithm,the image matching performance of the algorithm is improved in the case of large view point difference.Based on the research of above key technologies,a set of AR panoramic linkage monitoring system is built.The system uses augmented reality technology to mark the virtual information of buildings,roads and key targets in the surveillance images,the camera in high position is used to overlook the panorama of the area,and the distributed spherical camera in the scene is called through the linkage of the annotation information to shoot the area of interest from multiple angles actively,and the monitoring pictures of different cameras are displayed together to master the overall monitoring situation.The system can provide users with more intuitive and richer monitoring information,which has certain innovation and practical value.