多视结构光点云的自动无缝拼接

Close Multi-view Metrical Data Registration

  • 摘要: 基于视觉的单视点点云数据对纹理贫乏的工业钣金件多视角测量数据进行拼接。在三维量测系统的整体拼接中,旋转平台的转动使得每幅影像上的特征线在移动,因而影像与影像之间不存在真正意义上的同名特征线,以致所解算出的各个模型的点云之间不存在真正意义上的同名点。结合摄影测量学中基于闭合条件的独立模型法平差方法以及计算机视觉界的迭代最邻近点(iterative closest points,ICP)算法,对多视角测量数据进行拼接,实验结果验证了该算法的鲁棒性和有效性。

     

    Abstract: According to single-view point cloud,we presented a multi-view metrical data registration method for the industrial sheetmetal part without texture.In the whole registration,the actual corresponding feature lines don't exist between the images as the feature line in every image is shifted because of the turning of rotational table.Therefore no actual corresponding points exist between the points cloud,which is computed for every model.The multi-view metrical data registration method was described based on the combination of the close absolute model network adjustment and iterative closest point algorithm.Experimental results show that the method is robust and valid.

     

/

返回文章
返回