随着三维点云技术在工程测量、建模等方面应用的不断深入，对点云配准技术提出更高的要求。因此从实际应用的角度出发，对点云数据自动配准算法进行研究，针对常用的粗配准方法的不足，将尺度不变特征转换（SIFT）算法应用于三维点云粗配准中，研究了基于SIFT关键点的粗配准方法；针对传统的迭代最近点（ICP）算法存在的问题，将迭代系数和法线特征约束引入传统的ICP算法中提高配准的效率，实现了对传统ICP精配准算法的改进。实际应用结果表明，基于SIFT点的改进ICP点云自动配准算法可以有效提高配准效率和精度，具有较好的应用价值。

With the continuous deepening of 3D point application technology in engineering measurement, modeling, and other fields, higher requirements are put forward for point cloud registration technology. We studied the automatic registration algorithm for point cloud data from the perspective of practical applications. In response to the shortcomings of commonly used coarse registration methods, we applied the scale invariant feature transform(SIFT) algorithm to coarse registration, and studied a coarse registration method based on SIFT key points. In response to the problems existing in the traditional iterative closest point(ICP) algorithm, we introduced the iteration coefficients and normal feature constraints into the traditional ICP algorithm to improve the efficiency of registration, and achieved improvements to the traditional ICP precision registration algorithm. The practical application results show that the improved ICP point cloud automatic registration algorithm based on SIFT points can effectively improve registration efficiency and accuracy, and has good application value.