Study On Critical Technique In Video Based Virtual Environment

In complex virtual environments, geometry-based objects representation is always limited to the rendering complexity. Image-based modeling and rendering can greatly improve the realism and efficiency of virtual environment rendering. The dissertation discusses video-based modeling and rendering technology, extends the concept of image-based technology, simplifies the procedure of acquiring data in image-based technology, and avoids the difficult problem-image matching in computer vision. Constructing and modeling virtual environment through videos can improve the efficiency of modeling and enhance the virtual environment realism.In this dissertation, we first study the key technology of camera calibration in the non-laboratory circumstance, and propose a two-step method of cameara calibration, and give some problems detail analysis. Then, we discuss ideal camera motion, and track camera motion with extened kalman filter. Next, we simplify the camera model to affine projection, and apply factorization method in computer vision to get initial world model. Finally, we set up a prototype system to verify the ideas of video-based modeling. The original contributions of this dissertation include the following:By study of the noise in the camera calibration, this dissertation proposes the corresponding solutions. The proposed two-step camera calibration algorithm can minimize the effect of noise in camera calibration.Based on force-free camera motion model, extended kalman filter is used to track camera motion and eliminate noise.By analyzing the noise in the procedure of data acquiring and handling, the dissertation try to minize noise influence, improve the algorithm precision and propose some advices for better quality of virual environment.A prototype system is developed; the key technologies are verified and applied in projects.Summarily, this dissertation proposes the fundamental theory, model and method of video-based technology, and implements the video-based rendering in virtual environment. Such research systematizes and enriches the image-based object theory, and gives some solutions for stereo vision and 3D reconstruction.