Based On The Lcd Of The Camera Calibration And Three-dimensional Reconstruction Study

3D reconstruction of objects based on 2D images is an important topic in the field of computer vision. It has a broad application prospects in many fields such as virtual reality, 3D measurement, robot self-navigation, industrial inspection etc. The process of 3D reconstruction mainly involves image pre-processing, camera calibration, feature extraction and matching and 3D reconstruction algorithm.An improved sub-pixel edge contour detection algorithm based on orthogonal Fourier-Mellin moments is proposed. Canny operator is firstly used to extract pixel level image edges, then orthogonal Fourier-Mellin moments operator is used to determine sub-pixel image edges at the position of the pixel level image edges. The least-squares ellipse fitting method is used to the sub-pixel image edges to get more accurate sub-pixel ellipse center coordinates, and further improve the precision of camera calibration. Experiments proved that the sub-pixel detection algorithm has short computing time and high detection accuracy.It is difficult and costly to ensure the strict flatness of the calibration panel. In response to this issue, a nonlinear camera calibration method based on Liquid Crystal Display is proposed. Firstly, a calibration table, which has some circular feature points of different sizes, is designed and displayed on the LCD screen. Then several images of the calibration table are shot by rotating the camera around the optical axis. Finally camera parameters are calculated based on Zhengyou Zhang's calibration algorithm and the tangential distortion are considered. Experiments proved that the calibration method is simple and convenient and it has high calibration accuracy.A 3D reconstruction algorithm based on marker points using multiple images is studied after the camera parameters are obtained. The coded targets and uncoded targets are the marker points and are distributed on the surface of the target object. As each coded target has a unique identity, they can be used to compute poses of the cameras and reconstruct the centers of the coded targets; then the correspondence of uncoded targets can be estimated based on the theory of epipolar geometry; finally the centers of the uncoded targets can be constructed using the triangle reconstruction algorithm. Experiments have proved the feasibility of the reconstruction method.