Hovorov Viktor, Lalor Michael, Burton David, Lilley Francis
General Engineering Research Institute, Liverpool John Moores University, City Campus, Room 114, James Parsons Building, Byrom Street, Liverpool L3 3AF, UK.
Appl Opt. 2010 Mar 10;49(8):1459-71. doi: 10.1364/AO.49.001459.
Three-dimensional (3D) surface shape measurement is a vital component in many industrial processes. The subject has developed significantly over recent years and a number of mainly noncontact techniques now exist for surface measurement, exhibiting varying levels of maturity. Within the larger group of 3D measurement techniques, one of the most promising approaches is provided by those methods that are based upon fringe analysis. Current techniques mainly focus on the measurement of small and medium-scale objects, while work on the measurement of larger objects is not so well developed. One potential solution for the measurement of large objects that has been proposed by various researchers is the concept of performing multipanel measurement and the system proposed here uses this basic approach, but in a flexible form of a single moveable sensor head that would be cost effective for measuring very large objects. Most practical surface measurement techniques require the inclusion of a calibration stage to ensure accurate measurements. In the case of fringe analysis techniques, phase-to-height calibration is required, which includes the use of phase-to-height models. Most existing models (both analytical and empirical) are intended to be used in a static measurement mode, which means that, typically, a single calibration is performed prior to multiple measurements being made using an unvarying system geometry. However, multipanel measurement strategies do not necessarily keep the measurement system geometry constant and thus require dynamic recalibration. To solve the problem of dynamic recalibration, we propose a class of models called hybrid models. These hybrid models inherit the basic form of analytical models, but their coefficients are obtained in an empirical manner. The paper also discusses issues associated with all phase-to-height models used in fringe analysis that have a quotient form, identifying points of uncertainty and regions of distortion as issues affecting accuracy in phase maps produced in this manner.
三维(3D)表面形状测量是许多工业过程中的重要组成部分。近年来,该领域有了显著发展,目前存在多种主要的非接触式表面测量技术,其成熟程度各不相同。在众多3D测量技术中,基于条纹分析的方法是最有前景的方法之一。当前技术主要集中在中小型物体的测量上,而对于大型物体测量的研究还不够完善。众多研究人员提出的一种测量大型物体的潜在解决方案是多面板测量概念,这里提出的系统采用了这种基本方法,但采用了单个可移动传感器头的灵活形式,这对于测量非常大的物体具有成本效益。大多数实际的表面测量技术都需要包含一个校准阶段以确保测量准确。对于条纹分析技术,需要进行相位到高度的校准,这包括使用相位到高度模型。大多数现有模型(分析模型和经验模型)旨在用于静态测量模式,这意味着通常在使用不变的系统几何结构进行多次测量之前进行一次校准。然而,多面板测量策略不一定能保持测量系统几何结构不变,因此需要动态重新校准。为了解决动态重新校准问题,我们提出了一类称为混合模型的模型。这些混合模型继承了分析模型的基本形式,但其系数是以经验方式获得的。本文还讨论了与条纹分析中所有具有商形式的相位到高度模型相关的问题,识别出不确定性点和失真区域,这些是影响以这种方式生成的相位图准确性的问题。