二维点轮廓与矢量轮廓配准研究

Registration of Planar Point Cloud and Vector Graphics

在平面类零件的光学测量中,二维点轮廓与矢量轮廓的配准是关键算法,配准精度直接影响测量精度。针对平面类零件的配准问题,提出了基于形状特征函数的粗配准算法和二维矢量最近点迭代(ICP)精配准算法。利用角度距离图法将矢量图形的几何信息转化为独立于坐标系的连续函数,进而实现粗配准算法。基于平面上点与曲线的最近距离算法计算配准目标函数,给出了不同于传统的ICP算法的直接求解目标函数的解析方法,有效提高了算法效率。利用实例验证分析了该算法的高效性和可靠性。

In the optical measuring of planar parts, the registration of 2D point contour and planar vector contour is the key algorithm, the speed and precision of registration has a major impact on the speed and precision of measure result. In order to solve the problem of registration of 2D point cloud, a coarse registration algorithm based on shape feature function and a 2D iterated closest point（ICP） fine registration algorithm are researched. Through the angle-distance graph, geometry information of point contour and planar vector contour is translated to continuous function that is independent of coordinate system. The objective function of registration is calculated on account of nearest distance algorithm on planar point and curve, and analytic method of solving the objective function directly is given, which is different from traditional ICP algorithm. The efficiency of algorithm is improved significantly. Examples are exhibited to analysis the efficiency and reliability of the algorithm.