A Progressive Global Point Cloud Registration Method for Large Aircraft Exterior Measurement

Large aircraft impose stringent requirements on performance, structural stability, and reliability, and high-precision 3D measurement of the complete aircraft exterior is critical for manufacturing quality and assembly accuracy. To address insufficient flexibility, automation, and global registration accuracy, a flexi-ble modular high-precision automated 3D measurement method is presented. The method first acquires dense point-cloud data through partitioned scanning; then performs three-stage progressive registration, including coarse alignment, Predator++-based fine registration, and ICP refinement; finally, it is evaluated on real aircraft data and the 3DMatch and 3DLoMatch benchmarks. Multiview experiments achieve an MRE of 1.71°, an MTE of 0.12 mm, an RR of 98.1%, and an average runtime of 1.5 h; pairwise registration on 3DMatch reaches IR/FMR/RR values of 64.6%, 98.4%, and 90.0%, improving over SMVR by 4.1, 0.2, and 0.4 percentage points; on 3DLoMatch, IR/FMR/RR reach 30.3%, 84.1%, and 68.2%, with RR improving over SMVR by 0.3 percentage points.