Splicing And Dynamic Scheduling Of Massive Video In 3DGIS

With the deepening application of digital technology in the field of security,the integration of 3D-GIS and surveillance video systems has become a hot research issue.The fusion of video images into 3D geographic system can not only locate the geographical location of video,but also solve the problems of separate distribution,poor correlation and low utilization of traditional video surveillance.3D video fusion realizes the global observation of large scenes.Meanwhile,the temporal information is added to 3D virtual space.However,the current research methods seldom consider the spatial information of video and 3D scene fusion,resulting in poor fusion effect.At present,non-default rendering pipeline was adopted in most 3D video fusion methods,which leads to the more complex generation of texture coordinates.The corresponding GPU and rendering time take up more,resulting in the limited number of loaded videos.Therefore,it cannot meet the needs of dynamic management of video surveillance in large-scale scenes.Directed to the current research conditions of the limited number of videos and poor splicing fusion effect,this thesis optimized multi-channel video scheduling,splicing video and scene 3D model fusion methods.The research content mainly includes the following aspects.(1)The fusion scheduling method of multi-channel video and 3D scene is studied.Aiming at the problem of limited number of multi-videos,a dynamic scheduling method of massive videos based on octree index pixel mode LOD is proposed to load massive videos in terrain scene.In this process,in order to further improve the rendering speed,file mapping is used to store a large number of video nodes.In addition,pixel mode LOD model is used to eliminate the camera view cone to reduce the amount of rendering data,so as to better improve the video fluency.Dynamic scheduling of 70 videos in large-scale scenarios is realized.(2)The mosaic fusion method of video and 3D model is studied.Aiming at the problem of poor mosaic fusion effect,method is proposed to improve the mosaic fusion effect of video and 3D model by using camera internal and external parameter mapping,monitoring camera attitude information,camera real-time monitoring video stream decoding and visualization.Compared with the direct projection method,the results show that this fusion method is better than the direct projection method.