GENERAL POINT CLOUD FEATURE EXTRACTION AND MATCHING ALGORITHEM

3D scene perception and reconstruction based on laser point cloud is an important research direction in digital city, virtual reality and other fields. The common point cloud matching method generally refers to solving the point cloud correspondence under low dynamic conditions, but it cannot solve the point cloud matching problem under large dynamic conditions. Therefore, this paper proposes a new point cloud feature extraction and matching method, which is compatible with point cloud matching under low dynamic and large dynamic conditions. Experiments show that the proposed algorithm can adaptively solve the point cloud matching problem under various dynamic conditions, and has high accuracy.