摘要
针对传统编码结构光多频时间相位展开算法在实际三维测量中,由于不可忽略的噪声和其他误差源所导致的条纹阶次错误,严重影响三维测量精度的现象,提出一种基于深度学习的多频绝对相位展开算法。通过复杂曲面模型三维重建实验验证了该算法的可行性,实验结果表明在不同程度高斯噪声影响下,该算法三维重建结果相较于传统多频外差算法测量精度更高。
Aiming at the phenomenon that fringe order errors caused by non-negligible noise and other error sources seriously affect the accuracy of 3D measurement in the actual 3D measurement of traditional coded structured light multi-frequency temporal phase unwrapping algorithm,a multi-frequency absolute phase unwrapping algorithm based on deep learning was proposed.The feasibility of the proposed algorithm was verified by the 3D reconstruction experiments of complex surface models.The experimental results show that the 3D reconstruction results of the proposed algorithm are more accurate than the traditional multi-frequency heterodyne algorithm under the influence of different degrees of Gaussian noise.
作者
何群
薛林
张德健
王豪
王云森
HE Qun;XUE Lin;ZHANG De-jian;WANG Hao;WANG Yun-sen(School of Mechanical Engineering,Dalian University of Technology,Dalian 116024,China)
出处
《仪表技术与传感器》
CSCD
北大核心
2023年第4期93-96,101,共5页
Instrument Technique and Sensor
基金
国家自然科学基金项目(51905074)
辽宁省自然科学基金项目(2019-KF-04-04)。
关键词
结构光
相位测量
时间相位展开
三维重建
包裹相位
绝对相位
structured light
phase measurement
temporal phase unwrapping
3D reconstruction
wrapped phase
absolute phase
