一种新的基于结构光的三维视觉系统标定方法

在视觉系统的标定中，确定三维空间点与图象上对应点所组成的样本点对的坐标是问题的关键，为提高样本点的定位精度，通常采用棋盘格形状的图案，取其中的交点作为样本点．但在应用平面结构光的三维视觉系统中，样本点又必须在结构光平面中，无法应用上述方法，使得样本点的定位很困难，为了解决这一问题，我们提出了一种新的标定方法，该方法利用交比不变性原理，借助一块特殊设计的标定块，可以方便准确地确定样本点的位置．在此基础上完成了三维视觉系统的标定，文章给出了实验结果和标定精度，以及进一步提高精度的措施．

A NEW METHOD OF CALIBRATION IN 3D VISION SYSTEM BASED ON STRUCTURE-LIGHT

It is vital to the calibration of vision systems to determine the coordinates of sample pairs of object points in 3D spare and their corresponding points in image plane. In order to improve the locating accuracy of the sample points, it is a common practice to utilize chessboard pattern, and use the intersection points as the sample points. As for the 3D vision system based on plane structure-light the sample points are required to be in the structure-light plane. So the above method cpu not be used. To solve this problem, we present a new calibration method whose theoretical basis is the cross-ratio invariance, and the method uses a specially designed calibration etalon block. Using this method, it is simple to determine the location of the sample points explicitly, based on these location data, we can finish the calibration of the vision system. The paper gives the experimental results and the calibration accuracy, and also presents the measures to improve the accuracy.