面向建筑物重建的相机标定方法研究

由于灭点具有很多独特的几何属性且大量地存在于建筑物场景中，因此针对建筑物重建，提出了一种基于灭点的相机标定方法。首先通过构造各种几何约束关系，如灭点与相机矩阵间的关系、世界坐标原点和相机矩阵间的关系，逐步实现相机矩阵的度量重建和欧氏重建，获得相机矩阵的值，然后通过分解相机矩阵得到相机的内外部参数。该方法在标定过程中无须借助任何标定物，求解过程简单，标定速度快。实验结果表明，该方法可以满足虚拟现实中建筑物场景重建的需要。

Research on camera calibration used for building reconstruction

Vanishing point has many geometry attributes and exists abundantly in buildingscenes, based on this, a camera calibration method is proposed using vanishing points, which isespecially used for building reconstruction. Lots of geometry constraints are discovered, such asthe constraint relationship between vanishing points and camera matrix, the origin of worldcoordinate system and camera matrix. Then through matrix reconstruction and Euclidreconstruction, the camera matrix is calculated. Through decomposition of this matrix, the innerand outer parameters of the camera are achieved. In this method, no calibration object is needed;besides, it also has the advantage of simple calculation process, high calibration speed.Experiments show that it can meet the needs of building reconstruction in virtual reality system.