To address the problems that the Iterative Closest Point (ICP) registration algorithm tends to fall into local optimum and slow convergence of iterations, this paper proposes an improved ICP point cloud registration algorithm based on Intrinsic Shape Signatures (ISS) feature point combination, which firstly down samples the reference point cloud and the point cloud to be aligned with voxel grid. Then extracts the feature points by ISS algorithm and uses Fast Point Feature Histogram (FPFH) to characterize the feature points. After that, this paper finds the corresponding point pairs of the two sets of point cloud feature points, and then uses Random Sample Consensus (RANSAC) to remove the wrong corresponding point pairs. Finally, the two sets of point clouds with good initial poses are accurately aligned by the point-to-plane ICP algorithm based on Nanoflann acceleration, which further improves the registration accuracy and registration efficiency. The experimental results show that the algorithm reduces the number of iterations compared with the traditional ICP algorithm, and has a significant improvement in accuracy and speed; compared with the Scale Invariant Feature (SIFT) ICP algorithm, the Euclidean fitness score and registration time are reduced on average respectively 64.4% and 73.75%.