机器视觉中圆形特征的三维模型分析研究

 提出了一种机器视觉中圆形特征的三维模型分析方法。通过最小化数据点与所拟合椭圆的代数距离平方和，在满足二次曲线为椭圆的约束条件下，利用拉格朗日乘子法求得椭圆的5个特征参数。通过分析圆形特征的透视成像模型与原理，利用闭式解求得具有已知半径的圆形特征的三维位置和方向。对一组在标定模板上相同大小的圆进行特征检测与模型分析后，实验结果表明本文方法是有效和精确的。

3D model analysis of circular feature for machine vision

A novel 3D model analysis of circular feature is addressed in this paper for machine vision. We use an efficient method for fitting ellipses to data points by minimizing the algebraic distance subject to a constraint, which a conic should be an ellipse, and solve the ellipse parameters with Lagrange multiplier method. After analyzing the perspective model of circular feature, the 3D position and orientation of a circular feature with known radius is obtained using closed form solution. A set of identical circles are detected with the model analysis. Experimental results show that our method is efficient and accurate.