The Research On Auto Registration Method Of PTZ Camera In Augmented Virtual Environment

The network video surveillance is the third–generation video surveillance system,it's more powerful than the analogue and digital video surveillance system.This system can provide the video texture information for 3D GIS to build the Augmented Virtual Environment(AVE).While most people tend to be more concerned with the AVE of box camera,this paper try to discuss the video registration of Pan–Tilt–Zoom(PTZ)Camera in AVE.This registration of PTZ camera differs from that normal camera calibration method,because the parameters of the camera always changing.As far as the parameters are changing,automatic refreshing the pose of virtual camera in AVE is necessary.This paper tends to design the simpler video registration method for PTZ camera,the main contents of this method are:(1)Define the term ‘Virtual Image' and introduces its technical principles.Then describe the method of recovering 3D geographic coordinates of vertices from virtual image by analyzes the processing and display methods of vertices in the 3D scene graphic.Regarding the problem of video registration in 3D scene,this paper explored and applied the Shadow Map algorithm and described it's processing flow.(2)Describe the pose of virtual camera in 3D virtual environment with pose parameters and focal length parameter.Introduce the camera model based on the principle of pinhole imaging and summarizes the Direct Linear Transform(DLT)method to solve the camera matrix of camera model.This DLT method works under homogeneous linear equation generation and non-homogeneous linear equation generation.(3)According to the difference of virtual image imaging and camera image imaging,this paper explored and analyzed the feature detection method,feature description method and feature matching method base on points feature and line feature.With respect to generalized point photogrammetry,this paper showed a camera matrix solution model with the matching line segment features.(4)This paper try to show the different automatic registration method for PTZ camera.The local motion registration method describe the situation that the image changed a little after rotate the camera.While over–limit motion registration method describe the situation of variety image change after rotate the camera.(5)As discussed above,this paper uses a 3D GIS platform and related data to simulate the automatic registration of PZT camera in an augmented virtual environment.The experiment result prove the accuracy and feasibility of the methods and contributed to the further development of AVE.This paper solved the main problem of automatic registration of PTZ cameras in the augmented virtual environment.And provides related theoretical methods on how to feed back the changes of the camera's pose to the 3D virtual environment.This methods made contributed on developing 3D video GIS which will help to further enrich the application research scenarios of smart cities.