基于三角网格构建的扫描测头半径三维补偿

采用扩展的自组织特征映射神经网络探讨了三坐标测量机接触式密集数据采集的测头半径三维补偿。构建了基于三角形网格构建的测头半径三维补偿模型。经过训练，神经网络将整个数字化点群数据分成许多子区域，每个子区域用一个微切平面逼近；对子区域的分类核心，即神经元位置权重，沿微切平面法矢方向进行修正，得到逼近测头球心面的三角形网格II；根据微切平面的法线，对测头半径进行三维补偿，得到逼近接触曲面的三角形网格III。测头半径三维补偿的法矢方向，也可通过估算三角网格II顶点的法矢得到。算例表明所创建的测头半径三维补偿模型有效可行。

Three dimension radius compensation of touch probe based on reconstruction of triangle mesh

By using the Extended Self-Organizing Feature Map（ESOFM）neural network, an approach to the probe radius 3D（three dimension）compensation of the Coordinate Measuring Machine（CMM）for the dense 3D scattered data is developed based on the reconstruction of triangle mesh. After the neural network trained, the whole scattered data are divided into sub-regions whose classified core is represented by the weight vectors of the neurons. Every sub-region is approximated by a tangent plane. Then the neuron location weights of the probe radius 3D compensation model are adjusted along the normal vectors of the tangent plane, and triangle mesh II is reconstructed which draws near probe center surface. The probe radius is compensated according to the normal vector of the tangent plane, and triangle mesh III is reconstructed which draws near part surface. The normal vectors for probe radius 3D compensation can also be achived by computing vertex normal of triangular mesh II. The method is validated experimentally on three examples.