Aiming at the problem of broken and incomplete feature line extraction of 3D point cloud model, a feature line extraction algorithm based on reweighting of neighboring points is proposed. The algorithm is divided into two steps: extracting feature points and connecting feature points into lines. In the step of extracting feature points, the nearest neighbor reweighted local centroid operator is introduced to obtain the feature point set, and the feature line is constructed by the Euclidean minimum spanning tree. The experimental results show that compared with the traditional curvature-based algorithm, the nearest neighbor reweighted local centroid algorithm is more accurate and robust, and can effectively extract the geometric features of point cloud model.