Abstract:Aiming at the multi-scale feature detection of geometric models of aircraft parts, a robust multi-scale feature point detection algorithm is proposed. First, the algorithm designs an median filtering algorithm to obtain the accurate normals of the unstructured point cloud. Then, based on the calculated normals, calculates the local neighborhood fluctuations of each point to extract the initial feature points. Finally, for the problem of data redundancy of the initial feature points, a shrinkage optimization model is proposed to calculate the final feature data points in this paper. The experimental results show that the proposed algorithm is stable, simple and effective. The extracted feature point is complete, and it has better detection effect than traditional methods.