Research On Multi-scale Point Cloud Registration Method Based On A Non-cooperative Game

The three-dimensional(3D)point cloud is one of the most promising tools for representing and identifying 3D objects.In actual applications,to obtain the 3D point cloud model of the target,point cloud registration technology is required to rotate and translate the point clouds collected from different perspectives and stitch them into a complete point cloud.The critical step for registration is to find the appropriate feature descriptors.Two prevalent descriptors are global feature descriptor and local feature descriptor.The former represents the geometric and topological properties of the neighborhood in the entire 3D model,but it can not recognize the covered areas.The local descriptor focuses on narrow neighborhoods and the density of point clouds affects the extraction of local descriptor.In this paper,we present a multi-scale 3D point clouds based on registration method based on a non-cooperative game.By the combination of the curvature and eigenvalue variation,the key points are detected precisely under multiple scales.Furthermore,we develop a multi-scale covariance matrix descriptor to demonstrate local features of the key points.The multi-scale covariance matrix descriptor includes the geometric angles,dimensionality,the ratio of projection length and the difference of the curvature,which can describe the local geometric features of the key points more clearly and make feature descriptors more distinguished,especially for key points which are similar in a small range but are not similar in a large range.Besides,the matching method based on a bidirectional proportion strategy and the mismatched pairs deletion based on a non-cooperative game are used to find the optimal matching pairs.Compared with some popular registration methods,our method effectively reduces mismatch errors.Experiments show the efficiency and the robustness of the proposed algorithm for matching and registration of three-dimensional point clouds with Gaussian noise and deformed shapes.