针对ICP算法配准需要两点云有较好的初始位置否则无法获取准确匹配结果的问题,提出一种新的粗配准算法。调整两片部分重叠点云的初始位置;在求取一点处法向量的基础上,利用点云曲率信息,提取特征点,获取两点云每一特征点处的属性向量;通过相似度函数评价,寻找匹配特征点对进行粗配准。试验表明,该基于特征点提取和匹配的方法可为ICP算法提供良好的点云初始位置,并提高配准精度和可靠性。 In order to solve the problem that the registration of ICP algorithm needs a good initial position of two-point cloud and can not obtain accurate matching results, a new coarse registration algorithm is proposed. Adjust the initial position of two overlapping point cloud; based on the point vector, using the curvature information of point cloud to extract the feature points, get the attribute vector of each point of the two point clouds; through the similarity function evaluation , Looking for matching feature points for coarse registration. Experiments show that this method based on feature point extraction and matching can provide ICP algorithm with a good initial location of point cloud and improve registration accuracy and reliability.