高适应性单目线激光三维测量方法的研究
High Adaptability 3D Measurement Method Based on Line Laser Monocular Vision
姚宇 1张秋菊 1吕青 1焦露1
作者信息
- 1. 江南大学机械工程学院,江苏省食品先进制造装备技术重点实验室,江苏无锡 214122
- 折叠
摘要
在工业场景中有大量高动态范围表面的工件的三维测量需求,而现有的线激光单目视觉测量方法难以适应高动态范围测量场景的挑战.为保证高动态范围工件的三维测量精度和可靠性,提出了一种高适应性的线激光单目视觉测量方法.该方法改进了激光条纹区域分割、多尺度自适应激光条纹中心提取算法,提出了基于最大确定性推演的反光干扰的去除算法.最后通过点云拼接、统计滤波器去除稀疏离群噪点和体素栅格滤波器对点云降采样处理,完成高动态范围表面的工件的可靠测量.实验结果表明,对多个高动态范围的标准量块进行三维测量,最大测量误差小于±0.088 mm;对代表性的高动态范围工件翻领成型器进行三维测量,结果准确地测得了工件的真实三维形貌.所提方法对测量对象具有高适应性,测量结果具有良好的可靠性和准确性,满足一般测量要求.
Abstract
In industrial settings,there is a significant demand for 3D measurement of workpieces with high dynamic range sur-faces.However,existing line laser monocular vision measurement methods struggle to meet the challenges of such high dynam-ic range scenes.In order to ensure the accuracy and reliability of 3D measurement of workpiece with high dynamic range,a high adaptability linear laser monocular vision measurement method is proposed.This method improves the laser fringe region seg-mentation,multi-scale adaptive laser fringe center extraction algorithm,and presents a reflection interference removal algorithm based on maximum deterministic inference.Finally,point cloud splicing,statistical filter to remove sparse outlier noise,and voxel raster filter to sampling point cloud drop processing are used to achieve reliable measurement of workpiece on high dy-namic range surface.The experimental results show that the maximum measurement error is less than±0.088 mm for the three dimensional measurement of several high dynamic range standard blocks.A representative high dynamic range workpiece lapel shaper was measured and the real 3D shape of the workpiece was accurately measured.The method proposed in this paper has high adaptability to the measurement object,the measurement results have good reliability and accuracy,and meet the gen-eral measurement requirements.
关键词
三维测量/高动态范围工件/单目线激光/条纹中心提取Key words
three-dimensional measurement/high dynamic range workpiece/line laser monocular vision/fringe center ex-traction引用本文复制引用
出版年
2024
NETL
NSTL
万方数据