Information Registration Method for Three-Dimensional Body Data Measurement of Pilots

For the application of 3D body data measurement of pilots, multiple cameras are usually utilized to collect 3D point cloud information of the whole body of pilots from different viewpoints to complete the measurement, in which the registration of point clouds becomes a crucial step in the measurement. Targeting the issues encountered in the conventional iterative closest point (ICP) algorithm, including its tendency to converge to local optimal solutions easily and its high reliance on initial estimates, a human body point cloud registration method based on ISS-3DSC (Intrinsic Shape Signatures - 3D Shape Context) features combined with improved ICP is proposed. In the coarse registration, this method utilizes Intrinsic Shape Signatures (ISS) to extract key points and then combines them with 3D shape context features (3DSC) to obtain feature descriptions of these key points. In the fine registration phase, we propose an optimized point-to-plane ICP (PP-ICP) algorithm using linear least squares optimization. The experimental results show that compared with ICP, FPFH+SAC-IA+ICP, the proposed approach exhibits higher registration accuracy and efficiency, enabling rapid completion of point cloud stitching.