Improving Material Property Measurement Using Multi- Camera Digital Image Correlation
- Created: Friday, 28 February 2014
- Grove City College, Grove City, Pennsylvania; Oakland University, Rochester, Michigan; and University of Maryland Eastern Shore, Princess Anne, Maryland
The automotive industry faces growing pressure for more efficient vehicles. This pressure comes from many sources including government regulations, decreasing natural resources, and consumer demand. As the pressure begins to rise, automakers have a growing need for better fuel, more efficient energy-generating processes, and lighter vehicles. In an effort to reduce the weight of their vehicles, automakers have shown an increasing interest in using aluminum, provided it can withstand the same deformations as steel.
A multi-camera Digital Image Cor relation (DIC) system is applied to measure the material properties of aluminum (5754) specimens. Such tests are usually done using 2D (onecamera) or 3D (two-camera) DIC systems. A multi-camera DIC system includes three or more cameras and inherits all the advantages of a conventional 3D DIC system (with two cameras) such as full-field measurement, high accuracy, and high speed.
Digital Image Correlation is a non-contact evaluation method that uses CCD or CMOS cameras to record and track the individual points of a test specimen. By using two or more cameras, the software is able to calculate X, Y, and Z coordinates, as well as the strain, in each direction. This is accomplished by tracking the motion of pixels and small areas around them called subsets; the exact number of subsets will vary with different setups. The theory behind this method is that although a given area, or subset, may deform, the amount of light reflected from that area should remain the same. In this way, the software is able to track each point throughout the entire test regardless of the extremity of the deformation. In experimental conditions, the pattern matching and tracking are more complicated than described here.
Possible illumination changes and many other factors (e. g. the deformation of subsets) are taken into consideration. Conventionally, a 3D measurement with DIC is done with two synchronized cameras because, for a single spatial point, two cameras can provide enough data for 3D reconstruction. In this work, a three-camera DIC system is employed. This system follows the same fundamental procedure as a conventional 3D DIC system, but is capable of achieving higher accuracy and stability by involving redundant data captured by a third camera.
When three cameras are focused on the same area, any of them could act like a reference camera. The system is an equivalent of three independent DIC systems measuring the same area. For each point on the object's surface, there are three sets of data, and the best data among the three sets will be selected as the result. For a regular DIC system, the correlation error threshold is about 0.2 pixels. But the multisensory DIC system can easily reduce the error threshold to 0.05 pixels.
With a conventional 3D DIC system, data can only be correlated between two images. This requires the system to be very carefully calibrated in order to ensure that every image from one camera corresponds well to the same image from the second camera. If there are errors in the calibration, the data either cannot be correlated, or will correlate poorly. By adding a third camera, the number of correlations increases from one (camera 1 to 2) to three (camera 1 to 2, 1 to 3, and 2 to 3).
Since the correlation only requires two images, some of the data from each camera is redundant and will not be used. However, at some point in the correlation, images from all three cameras are used, so each camera is required. In this way, the DIC correlation is less dependent on the calibration, and therefore more user-friendly and more likely to yield a good correlation. These advantages also make multi-camera DIC very advantageous when testing a large number of samples, as the system does not have to be recalibrated nearly as often as a 2D or 3D DIC system.
One major problem in DIC tests on aluminum is that slipping or peeling off usually occurs when the specimen is subjected to large strain. In this work, an Iwata Revolution CR airbrush is used to apply the speckles, and the thickness of the paint layer is greatly reduced.
The experimental setup for this project consisted of three CCD cameras, a green LED light source, a tensile machine, a solid background, and several computers for recording data. The figure shows the setup in detail. The dark background and a green LED array were used to provide the best illumination of the test specimen, and to reduce background noise, since the entire background would appear to the DIC system as an area with a gray value of 0.
This work mainly shows the capability of a multi-camera DIC system that is upgraded from a conventional 3D DIC system to measure multiple material properties of aluminum specimens. Compared to most steels, aluminum is soft and has better ductility, so the measurement range of the system should cover most of the common materials used in the industry. The multi-camera DIC system makes it easier to reach higher evaluation accuracy. The most important material properties could be exported directly from the DIC results.
This work was done by Caleb P. Chovan of Grove City College, Betelhem Mengiste of the University of Maryland Eastern Shore, and Xu Chen, Lianxiang Yang, and Laila Guessous of Oakland University. The full technical paper on this technology is available for purchase through SAE International at http://papers.sae.org/2013-01-1428/.