光学精密工程2025，Vol.33Issue(9)：1396-1406,11.DOI:10.37188/OPE.20253309.1396
基于改进ICP算法的大尺寸焊接构件三维形貌测量
Improved ICP algorithm-based 3D morphology measurement of large-size welded components
摘要
Abstract
A three-dimensional topography measurement method for large-scale components is proposed to achieve high-precision and high-efficiency measurement.To address the issues of low accuracy and poor stability inherent in binocular structured light measurements,a complementary positive and negative Gray phase-shifted structured light coding technique is developed.This approach mitigates the susceptibility of binocular structured light systems to systematic errors and inaccurate phase unwrapping during actual mea-surement.The method employs a robust pixel classification technique for binarizing the positive and nega-tive Gray codes,alongside an improved Gaussian filtering algorithm applied to the phase-shift code imag-es.To improve the measurement accuracy of existing point cloud,an improved iterative nearest point cloud stitching algorithm is introduced.This algorithm extracts and filters point clouds from overlapping re-gions of adjacent perspectives to eliminate interference from non-overlapping areas and integrates the Lev-enberg-Marquardt optimization algorithm into the iterative process to enhance robustness against initial po-sition sensitivity and noise interference.Compared to traditional iterative nearest point stitching methods,the proposed algorithm increases accuracy by 55%,accelerates stitching efficiency by several times,and reduces iteration count by 61%.Experimental results demonstrate that the developed system achieves length measurement accuracy better than 450 μm/m at a measurement distance of 700 mm and a camera angle of 65°,with point cloud stitching computation times of approximately 25 ms between adjacent views.These findings confirm that the method satisfies the accuracy and efficiency requirements for three-dimensional topography measurement of large-scale welded components.关键词
三维形貌测量/双目结构光/鲁棒像素分类法/改进迭代最近点/正反互补格雷码Key words
three-dimensional morphorogy measurement/binocular structured light/robust pixel classifi-cation/improved ICP algorithm/positive-negative gray phase shift分类
信息技术与安全科学引用本文复制引用
蔡引娣,王宇轩,牛超,朱祥龙,康仁科..基于改进ICP算法的大尺寸焊接构件三维形貌测量[J].光学精密工程,2025,33(9):1396-1406,11.基金项目
国家重点研发计划资助项目(No.2022YFB4600900) （No.2022YFB4600900）
国家自然科学基金联合基金集成项目(No.U23B6005) （No.U23B6005）
中央高校基本科研业务费(No.DUT24MS005) （No.DUT24MS005）
高性能精密制造全国重点实验室开放课题资助(No.ZY202408) （No.ZY202408）
