中文摘要：

ICP算法是三维点云数据精确拼接过程中的主流算法。文章对目前国内外ICP及其各种改进算法的发展现状进行了系统地分析与研究。将ICP算法分为4个主要阶段：（1）对原始点云数据进行采样；（2）确定初始对应点集；（3）去除错误对应点对；（4）坐标变换的求解。分别对各个阶段中典型方法的基本思想和关键技术进行了分类与分析，并在精度与效率等方面对这些算法进行了比较。最后对目前算法研究中的难点问题及未来的研究重点进行了展望。

英文摘要：

The ICP algorithm has become the dominant method in fine registration of 3D point clouds. A comprehensive survey on the ICP algorithm and its variants is presented in the paper. The procedure of existing ICP and its variants is divided into four steps. （1） sampling the source points, （2） establishing the original correspondences, （3） removing the unreliable correspondences, （4） estimating the transformation parameters. In each step, the basic ideas and key techniques of various typical methods are classified and analyzed, and their advantages and shortcomings are studied and compared with respect to accuracy and efficiency. Finally, some open issues and relevant solutions are presented, and some discussion on the future prospects are provided.