This paper introduces a point cloud feature point detection method based on a hierarchical clustering algorithm to address the limitations of traditional methods in accurately detecting detailed features and reflecting the true object information. The minimum spanning tree and depth first search algorithm are used to adjust the direction of the normal vector of each triangle formed by each point and its neighborhood points. Non feature points and candidate feature points in the point cloud model are detected by Gaussian mapping of normal vector. For candidate feature points, hierarchical clustering algorithm is used to judge whether they are feature points. Experimental results demonstrate the effectiveness of the proposed algorithm in accurately detecting feature points within scattered point cloud data, including those with unclear details. Specifically, the method detected 810, 933, 2955, and 3941 feature points for the Sheep, Fandisk, Bunny, and Dragon point cloud models, respectively, surpassing the performance of other feature point detection methods.