Accuracy Analysis of an Oblique Underwater Laser Lightsheet Triangulation System

Accuracy Analysis of an Oblique Underwater Laser Lightsheet Triangulation System
复制标题

DOI:
10.1007/s41064-022-00196-x
复制
发表时间:
2022-02-22
影响因子:
4.1
通讯作者:
Maas, Hans-Gerd
Maas, Hans-Gerd
中科院分区:
地球科学4区
文献类型:
--
作者:
Sardemann, Hannes;Mulsow, Christian;Maas, Hans-Gerd

文献摘要

被引文献

相似文献

激光光片三角测量是一种成熟的光学测量方法,在工业应用中得到了广泛的应用。经过一些调整,这项技术也可以在水下使用。将绿色激光线投影仪和摄像机放置在水密外壳内的固定底座上,可以以低成本灵活准确地进行水下测量。为了实现物体表面的近正交相交,需要将相机和激光器斜置于外壳界面。当光通过空气-玻璃和玻璃-水界面传播时,在几何建模中必须严格考虑折射影响。本文提出的测量和校准方法利用了将光片分割成多个子光束的概念,这些子光束被跟踪,折射并与图像观测相交以接收三维测量。在标定步骤中,确定相机、激光器和接口之间的相对方向。在理论精度分析中,在深度为15 cm的范围内,估计校准参数引起的误差影响的标准偏差为0.2至0.4 mm。将三角测量传感器原型应用于两个已知几何形状的测试对象的实际测量。在平稳单次扫描测量中验证了预测的准确性。此外,传感器沿着测试对象移动,使用六自由度方法确定其外部方向。与参考相比,多次单次扫描的组合结果是密集的点云,标准偏差为0.3 mm。
Laser lightsheet triangulation is a well-established optical measurement method, which is frequently used in industrial applications. With some adaptions, the technique can also be used underwater. Placing a green laser line projector and a camera at a fixed base inside a watertight housing enables flexible and accurate underwater measurements at low cost. To achieve near-orthogonal intersections on the object surface, camera and laser need to be placed oblique to the housing interface. Refraction influences have to be considered strictly in geometric modelling, with the light propagating through the air-glass and glass-water interfaces. The measurement and calibration methods presented in the paper utilize a concept of splitting the lightsheet into multiple sub-beams, which are traced, refracted, and intersected with image observations to receive 3D measurements. In a calibration step, the relative orientation between camera, laser and interfaces is determined. In a theoretical accuracy analysis, the error influences caused by the calibrated parameters is estimated with a standard deviation of 0.2 to 0.4 mm in a depth range of up to 15 cm. A prototype triangulation sensor is applied for practical measurements of two test objects with known geometry. The predicted accuracy is validated in stationary single scan measurements. The sensor is furthermore moved along the test objects, using a six-degrees-of-freedom method to determine its exterior orientation. The combination of multiple single scans results in dense point clouds with 0.3 mm standard deviation compared to a reference.