逆向工程中回转点云的配准研究

Registration in Reverse Engineering on Revolving Point Clouds

针对大量回转型点云配准效率不够高的问题,提出一种先利用该类点云的回转半径特征搜索层差,再利用配合角度进行校核计算的方法。对于2块回转形状的点云,其在同一高度的回转半径是近似相同的;另外,如果他们能很好地进行配合,那么存在2条夹着配合断裂线的回转体母线。在同一高度上,这2条母线到断裂处弧线所对应的圆心角之和为一个近似的定值。正是利用回转点云的这2个特征,可以快速判断2块点云是否可以配合。在此基础上,为了进一步提高效率,在搜索层差的过程中,只是选取部分点云层进行搜索。文章主要通过三坐标扫描仪获取机械零件的外表面形状数据,然后进行配合验证,实验结果验证了该方法的可行性。该配准方法相对原先方法可明显提高效率。

For the low registration efficiency of a large number of rotary point clouds,the article proposed a registration approach： searching the layer difference between two point clouds using the radius first,and then checking the registration angles. For two rotary point clouds,they had approximative radius at the same height,at the same time,if they could be matched well,there would be two bus bars at opposite side of the break line. At the same height,the adding angle of two points＇ registration angles was a constant value which could be used to estimate if they could be matched correctly. In order to improve the efficiency,we could merely choose part of the radius for matching. Using the CMM scanners and getting the outer surface point clouds for experimental verification,the algorithm in this article had been proved perfect.