Simulation System Of Camera Characteristics Transform Based On Shooting Images

In the aerial remote sensing technique, images based on kinds of imaging condition are required. In fact, because of imaging condition limiting, some image sequence are only got. Therefore, if current camera's shooting state, such as shooting position or posture and parameters such as focal length and filed of view can not meet the demand, simulation which according to designed algorithm based on shooting images sequence generates images is necessary.The simulation system uses OpenSceneGraph software programming to achieve airborne camera characteristics transformation based on shooting images in Visual C++2008 platform. The system main function is to generate images when camera's position, attitude angle and internal parameters make changes, and to give simulation interface and simulation results. Finally the correctness of subject research has been tested.The main contents include the following:1) Research on Technology of Computer Imaging based on OpenGLAt first the issue analyzes imaging mechanism of OpenGL and linear camera respectively, then the correlation between OpenGL imaging and photogrammetry is determined. Based on the correspondence between the two imaging parameters by comparison, Computer imaging under specific condition is eventually determind by use of OpenGL.2) Research on Model of Camera Motion TransformationThe measurement model of camera motion includes translational and rotational movement pattern, and their combination. This system achieve images transformation based on airborne camera modle.The priciple is that reference image is generated by using of original image and then simulated image is created based on motion transformation algorithm.3) Imaging Performance of Camera ParametersSince the program using OSG is much better than OpenGL on efficiency of scene development and rendering, this simulation system uses OSG software programming to get images when camera parameters make changes, which is more efficient.4) Technology of Panorama StitchingIn order not to produce blank image message when transform, this system should make panorama masaic before image transform to get complete panorama as much as possible. This paper proposes a algorithm for panorama masaic based on feature points. Feature points are firstly extracted by Harris corner detection, then feature points matching according to the relationship among points is realized, and smooth fusion method is used to blend images.