3D Reconstruction And Object Tracking Based On Stereovision

In this paper, the key techniques in 3D reconstruction and object tracking are studied in detail, which is based on the theory of binocular stereovision. The developed algorithms are partly implemented with Visual C++ .NET.In order to be used in Surgical Navigation System of Computer-Assisted Surgery, the method for camera calibration is required to be simple and flexible. Through studying and analyzing several typical methods for camera calibration, Zhang's flexible new camera calibration method is chosen and used in our system. Stereo matching is a key technique in stereovision. Based on a thorough study of available bibliography, stereo matching is studied and discussed in detail. Simultaneously, the methods for 3D object location and reconstruction are analyzed, the technique of object tracking based on Kalman filter is studied in theory, and the mathematical model of 2D Kalman filter is proposed also. With full consideration of the requirements of the undergoing project, a new method for feature extraction and object tracking based on color information is proposed. At last, a system of camera calibration and 3D reconstruction is implemented in this paper, which is based on the function library of OpenCV. It is proved by the system that the method is feasible.