基于随机采样一致性算法的平面匹配方法

针对传统的平面图像标定点匹配算法计算量大、准确性不高等问题，提出一种基于随机采样一致性(RANSAC)算法的快速高精度的平面图像标定点匹配方法。该方法首先基于双圆锥曲线模型，借助于椭圆边缘点附近的梯度信息求取椭圆的切线，由切线集合拟合出椭圆参数，并利用这些参数计算出椭圆圆心，即图像上的标志点；然后用RANSAC随机采样算法对标定板上的点和图像上的点进行匹配。实验验证该方法实现起来简单并且有较高的准确率。

Matching method of plane image based on random sample consensus algorithm

Concerning the problems of large amount of calculation and poor accuracy of traditional matching method of plane image target point, a fast and highly precise matching method of plane image target point was proposed based on RANdom SAmple Consensus (RANSAC) random sampling algorithm. Firstly, based on the dual conic model, the ellipse's gradient vector was estimated by exploiting directly the raw gradient information in the neighborhood of an ellipse's boundary. Secondly, the ellipse's parameters and centers with the gradient vector field were matched. At last, the points in image with the target point in calibration board were matched with the help of RANSAC random sampling algorithm. The experimental results verify the method is simple and has a high degree of accuracy.