中文摘要：

多视数据最近迭代点（ICP）对齐广泛应用于反求工程及特征识别中,为了提高对齐精度以及稳定性,提出了一种基于移动最小二乘（MLS）曲面的ICP多视数据对齐方法.该方法的主要思想是直接利用MLS曲面表示原始曲面模型,将其他视角数据ICP对齐到该MLS曲面,其优点是无需对测量数据进行额外的去噪处理与数据分割,即可实现多视点云数据的精确对齐.实验结果表明,该方法能得到更加精确稳定的对齐结果.

英文摘要：

Multi-view point-data ICP registration is widely used in reverse engineering and feature recognition.A moving-least-squares（MLS） surface based ICP registration approach is proposed to improve the registration accuracy and stability.The main idea is to register the point-data on to the MLS surface which is employed to represent original surface model directly without any additional noise filtering or segmentation process.The experimental results show that more accurate and stable registration result can be obtained with the proposed approach.