中文摘要：

提出一种三维（3-D）视觉测量数据同步自动配准方法。利用改进的3-Dshape context描述符建立当前测量数据与已获得数据间的匹配点对,得到候选的重叠数据,并用最近点迭代（ICP）算法进行验证和精确配准。将视觉传感器每次获得的测量数据与已获得的测量数据配准,最终将所有测量数据统一到同一坐标系下。实验时,采用视觉传感器对维纳斯石膏像进行多角度测量,对测量数据进行同步配准,获得石膏像表面完整的3-D模型。该配准方法无需任何外部辅助设备,便于操作者观察测量进度并及时调整视觉传感器路径规划,从而提高三维形貌测量效率。

英文摘要：

A synchronous and automatic registration method for 3-D vision measurement is proposed, An improved 3-D shape context is developed to establish corresponding point pairs between current meas- urement data and previous ones, then the candidate overlapping measurement data are selected, and verified and accurately registered by iterative closest point algorithm. Each time the current measurement data from vision sensor is registered with the measurement data acquired previously, and finally all the measurement data can be unified at the same coordinate frame. In the experiment,a plaster statue of Venus is scanned by a vision sensor from different viewpoints. The scanned data are synchronously and automatically registered with the proposed method,and the 3-D model of the plaster statue is reconstructed, The proposed registration method needs no auxiliary equipments. It is convenient for operators to observe the measurement progress and adjust the path planning of scanner in time,and improves the efficiency of 3-D shape measurement.