Elsevier

Optics and Lasers in Engineering

Volume 98, November 2017, Pages 198-204
Optics and Lasers in Engineering

Multi-camera digital image correlation method with distributed fields of view

https://doi.org/10.1016/j.optlaseng.2017.05.003Get rights and content

Highlights

  • Multi-camera DIC system for large structures measurements has been presented.

  • The system has been used in in-situ measurements of a hall’s arch.

  • High accuracy data stitching has been obtained with laser tracker system.

Abstract

A multi-camera digital image correlation (DIC) method and system for measurements of large engineering objects with distributed, non-overlapping areas of interest are described. The data obtained with individual 3D DIC systems are stitched by an algorithm which utilizes the positions of fiducial markers determined simultaneously by Stereo-DIC units and laser tracker. The proposed calibration method enables reliable determination of transformations between local (3D DIC) and global coordinate systems. The applicability of the method was proven during in-situ measurements of a hall made of arch-shaped (18 m span) self-supporting metal-plates. The proposed method is highly recommended for 3D measurements of shape and displacements of large and complex engineering objects made from multiple directions and it provides the suitable accuracy of data for further advanced structural integrity analysis of such objects.

Introduction

The 3D digital image correlation (3D DIC) technique (also known as Stereo-DIC) is the well-established non-coherent light based method which provides full-field, non-contact measurements of shape, displacements and strains of engineering objects [1]. The 3D DIC method is a combination of 2D DIC (which uses a single camera and provides in-plane displacements) and stereo-vision provided by a pair of cameras. The 3D DIC is suitable for shape and deformation measurements of both - planar and curved objects and it can measure all three components of displacements of a test object. The method is characterized by scalable sensitivity and dimensions of a field of view and simple instrumentation. Since its origin in the early eighties [2] it has undergone many changes and modifications, which have improved computation efficiency and accuracy [3], [4], [5], [6], [7], adapted to new application fields [8], [9], [10], [11], [12], enhanced the hardware [13], [14], [15], provided solutions to work in extreme environments [16], [17] and enabled new measurement possibilities, such as 3D strain fields determination [18]) or performing measurements distributed in time [19]. Despite all this progress, 3D DIC, similarly to other machine vision based methods, faces the basic problem which is an interdependence of the accuracy and field of view (FoV) of a single 3D DIC system. It means that the accuracy of displacement measurement decreases with the increased FoV. The most frequent solution to this problem is extending a FoV through implementation of multiple cameras and further combining the data by means of a spatial data stitching (SDS) algorithm in which data obtained with individual systems are stitched together into a common coordinate system. A SDS algorithm has to be tailored to fit one of two measurement strategies, namely:

  • the strategy with overlapping areas of fields of view of multi-camera DIC,

  • the strategy with fields of view of multi-camera DIC distanced from each other and with no overlapping areas.

The first strategy, which is possible only when any two neighboring measurement systems (or all of them) have overlapping area within their fields of view, is described in papers [20], [21], [22], [23]. The calibration target in the overlapping area can be viewed by at least two systems at the same time. The knowledge of the 3D position of markers within calibration target in the local coordinate systems of 3D setups enables calculation of the geometrical transformation between coordinate systems of 3D setups. Specifically when big objects need to be measured it is not possible to assure common fields of view of all cameras (in contrast to [20] solution). In such cases, the cameras can be arranged in a wall-of-cameras configuration (for measurements of flat objects), or surrounding cameras (for cylindrical objects). Wang in [21] and Chen in [22] applied surrounding cameras configuration for measurements of cylindrical objects. In both papers the authors used each two neighboring cameras as individual 3D DIC system. Each system has been calibrated separately, but with sequential utilization of geometrical transformation between cameras it was possible to refer all data to the common coordinate system. In the presented solution it was required to have some overlaid area in the field of view of neighboring cameras. Somewhat different approach has been presented in [24] by Leblanc et. al. The Authors realized displacements and strains measurements of a wind turbine blade with a single, portable 3D DIC system. In this solution, 16 point clouds representing different areas of the object have been stitched together using a set of markers applied directly to the structure. The solution with portable 3D DIC system is restricted to an object which is changed in incremental way, while staying static through the entire measurements process.

Although the first measurement strategy extends significantly the ability of 3D DIC method to measure big objects, in many cases it is still insufficient. In industry, the crucial parts of installations can be distanced by tens (or even more) of meters and therefore the second measurement strategy with the distributed configuration of multi-camera DIC system (with no overlapping areas of fields of view) has to be used. When implemented, it is extremely useful for such important tasks as validation of numerical models of complex buildings [24] or civil engineering structures [25], as well as calibration of the numerical models of old installations (e.g. pipeline network in chemical industry), which do not operate in accordance to design assumptions due to many years of service. The first attempt to provide the measurement methodology in this scenario was presented by Malesa et al. in [23]. In this case, the data stitching procedure was performed with the use of external method, namely geodetic surveying. In order to determine transformation between individual 3D DIC setups, fiducial markers were applied to the object. Similarly to the previous strategy, 3D positions of all the markers need to be referenced to the common coordinate system. As none of the 3D DIC setups view all markers at the same time, determination of the markers positions was performed by geodetic surveying. The accuracy of data stitching procedure in this case is dependent on the quality of a standard camera calibration of individual 3D DIC setups and on the accuracy of the aiding method. Unfortunately, the utilization of geodetic surveying technique had not provided the sufficient accuracy of data stitching. In this paper, in order to overcome the limitations of the existing multi-camera 3D-DIC systems employed to displacement measurement for big engineering objects with areas of interests distributed in space, the authors proposed to employ a laser tracker [26] as the aiding method. During the measurements, the laser tracker follows (by means of an angular encoder and interferometry) the position of optical target (marker) with high accuracy. The laser tracker measures the position of markers in a global coordinate system, simultaneously 3D DIC systems measure the position of markers noticeable in their individual fields of view. Information on markers position in global and local coordinate systems, enable transformation of local coordinate systems of each 3D DIC system to common coordinate system. The high accuracy of the transformations has been validated by means of coordinate measuring machine in laboratory conditions. The effectiveness and performance of the proposed method and system are verified performing measurements and monitoring of a hall made of arch-shaped (18  m span) self-supporting metal-plates exposed to weather conditions.

Section snippets

General concept of measurements in distributed fields of view

The main challenge during the measurements of an object with distributed (non-overlapping) fields of view is connected with transforming many local coordination systems of a single 3D DIC setups into a single global coordinating system. The procedure is based on a standard geometric transformation between local coordinate systems (OX1Y1Z1,,OXnYnZn - related to each individual 3D DIC setup, where n indicates the number of 3D DIC system) and global coordinate system (OX0Y0Z0 - related to the

Validation of the accuracy of displacements measurements

The performed tests consider the validation of the accuracy of displacements measurements, which in multi-camera 3D DIC system is determined by the lowest accuracy of a component 3D DIC setup. Therefore, the laboratory tests were performed with utilization of a single 3D DIC setup. The obtained displacement measurements results were transformed into global coordination system (according to the proposed methodology) and compared to reference data. In laboratory conditions the field of view (FOV)

Measured object

The first implementation of the proposed method was performed during in-situ measurements of hall made of arch-shaped self-supporting metal-plates (Fig. 5a)), in cross-section of an arch, 8  m high and 18  m wide. Each arch-shaped metal plate has a cross section of a trough, the arches are jointed by double lock seam. This is a new type of buildings in civil engineering adopted directly from military applications, characterized by low cost and time of production. However, due to complex

Conclusions and future works

This paper presents the procedure that enables reliable determination of transformations between local (3D DIC) and global coordinate systems and therefore allow to perform measurements with a multiple 3D DIC systems with fields of view distributed in space. The proposed methodology was validated in laboratory conditions and implemented in measurements of large scale construction, the obtained data will support the process of updating the FEM model of this construction. The error of

Acknowledgments

The authors gratefully acknowledge financial support from the statutory funds of the Faculty of Mechatronics of Warsaw University of Technology and OPT4-BLACH project financed by the National Center for Research and Development http://www.ncbr.gov.pl/en/.

References (32)

  • M. Sutton et al.

    Image correlation for shape, motion and deformation measurements

    (2009)
  • W. Peters et al.

    Digital imaging techniques in experimental stress analysis

    Opt Eng

    (1982)
  • B. Pan et al.

    Study on subset size selection in digital image correlation for speckle patterns

    Opt Express

    (2008)
  • B. Pan et al.

    Fast, robust and accurate digital image correlation calculation without redundant computations

    Exp Mech

    (2013)
  • M.A. Sutton et al.

    The effect of out of plane motion on 2d and 3d digital image correlation measurements

    Opt Lasers Eng

    (2006)
  • K. Malowany et al.

    The application of 3d digital image correlation to track displacements and strains of canvas paintings exposed to relative humidity changes

    Appl Opt

    (2014)
  • Cited by (38)

    • Pressure-sensitive paint integrated with digital image correlation for instantaneous measurement on rotating blades

      2022, Aerospace Science and Technology
      Citation Excerpt :

      So, in this region, the angle between the normal vector of the surface and the image plane of the camera is too large and the measurement accuracy is reduced. This can be improved by properly adjusting the stereo camera position or using a multiple cameras system [45–47]. The dynamic point cloud data were measured at a fixed circumferential position.

    View all citing articles on Scopus
    View full text