Quality assessment of speckle patterns for digital image correlation

Digital image correlation (DIC) is an optical–numerical full-field displacement measuring technique, which is nowadays widely used in the domain of experimental mechanics. The technique is based on a comparison between pictures taken during loading of an object. For an optimal use of the method, the object of interest has to be covered with painted speckles. In the present paper, a comparison is made between three different speckle patterns originated by the same reference speckle pattern. A method is presented for the determination of the speckle size distribution of the speckle patterns, using image morphology. The images of the speckle patterns are numerically deformed based on a finite element simulation. Subsequently, the displacements are measured with DIC-software and compared to the imposed ones. It is shown that the size of the speckles combined with the size of the used pixel subset clearly influences the accuracy of the measured displacements.     

Strain field denoising for digital image correlation using a regularized cost-function

In digital image correlation (DIC), the widely used forward-additive Newton–Raphson (FA-NR) algorithm and the recently introduced equivalent but more efficient inverse-compositional Gauss–Newton (IC-GN) algorithm are capable of providing both displacements and displacement gradients (strains) for each calculation point. However, the obtained displacement gradients are seriously corrupted by various noises, and for this reason these directly computed strains are usually considered as useless information and therefore discarded. To extract strain distributions more accurately, much research efforts have been dedicated to how to smooth and differentiate the noisy displacement fields using appropriate numerical approaches. In this contribution, contrary to these existing strain estimation approaches, a novel and alternative strain estimation approach, based on denoising the noisy strain fields obtained by FA-NR or IC-GN algorithm using a regularized cost-function, is proposed. The effectiveness and practicality of the proposed strain estimation technique is carefully examined using both computer-simulated images with imposed homogeneous and inhomogeneous deformation, and experimentally obtained images. Experimental results reveal that the strains obtained by the proposed method are comparable to those determined by post-processing of the displacement fields using conventional pointwise least squares strain estimation approach.     

Two-step digital image correlation for micro-region measurement

A method of two-step digital image correlation is well developed with more stable and reliable calculating technology, which consists of a simple searching method and an iterative correlation method. This new method can not only improve the calculating speed and the measuring accuracy, but also simplify the process of the experiment. In order to further increase the sensitivity of the technique, the sub-pixel reconstruction is performed in sub-image by utilizing the higher precision calculation of bicubic spline interpolation value method, and the accuracy of displacement is extended to better than 0.01 pixel; the strain resolution is limited to less than 0.0002 in micro-region. The above method is applied to quantify the micro-deformation of bimaterial sample and coating sample. The experimental results show that the method of two-step digital image correlation is a potential boon to investigations at extremely small-size scales.     

A novel speckle pattern—Adaptive digital image correlation approach with robust strain calculation

Digital image correlation (DIC) has seen widespread acceptance and usage as a non-contact method for the determination of full-field displacements and strains in experimental mechanics. The advances of imaging hardware in the last decades led to high resolution and speed cameras being more affordable than in the past making large amounts of data image available for typical DIC experimental scenarios. The work presented in this paper is aimed at maximizing both the accuracy and speed of DIC methods when employed with such images. A low-level framework for speckle image partitioning which replaces regularly shaped blocks with image-adaptive cells in the displacement calculation is introduced. The Newton-Raphson DIC method is modified to use the image pixels of the cells and to perform adaptive regularization to increase the spatial consistency of the displacements. Furthermore, a novel robust framework for strain calculation based also on the Newton-Raphson algorithm is introduced. The proposed methods are evaluated in five experimental scenarios, out of which four use numerically deformed images and one uses real experimental data. Results indicate that, as the desired strain density increases, significant computational gains can be obtained while maintaining or improving accuracy and rigid-body rotation sensitivity.     

A fast digital image correlation method for deformation measurement

Fast and high-accuracy deformation analysis using digital image correlation (DIC) has been increasingly important and highly demanded in recent years. In literature, the DIC method using the Newton-Rapshon (NR) algorithm has been considered as a gold standard for accurate sub-pixel displacement tracking, as it is insensitive to the relative deformation and rotation of the target subset and thus provides highest sub-pixel registration accuracy and widest applicability. A significant drawback of conventional NR-algorithm-based DIC method, however, is its extremely huge computational expense. In this paper, a fast DIC method is proposed deformation measurement by effectively eliminating the repeating redundant calculations involved in the conventional NR-algorithm-based DIC method. Specifically, a reliability-guided displacement scanning strategy is employed to avoid time-consuming integer-pixel displacement searching for each calculation point, and a pre-computed global interpolation coefficient look-up table is utilized to entirely eliminate repetitive interpolation calculation at sub-pixel locations. With these two approaches, the proposed fast DIC method substantially increases the calculation efficiency of the traditional NR-algorithm-based DIC method. The performance of proposed fast DIC method is carefully tested on real experimental images using various calculation parameters. Results reveal that the computational speed of the present fast DIC is about 120-200 times faster than that of the traditional method, without any loss of its measurement accuracy     

Use of rigid-body motion for the investigation and estimation of the measurement errors related to digital image correlation technique

The aim of this work is to investigate the sources of errors related to digital image correlation (DIC) technique applied to strain measurements. The knowledge of such information is important before the measured kinematic fields can be exploited. After recalling the principle of DIC, some sources of errors related to this technique are listed. Both numerical and experimental tests, based on rigid-body motion, are proposed. These tests are simple and easy-to-implement. They permit to quickly assess the errors related to lighting, the optical lens (distortion), the CCD sensor, the out-of-plane displacement, the speckle pattern, the grid pitch, the size of the subset and the correlation algorithm. The errors sources that cannot be uncoupled were estimated by amplifying their contribution to the global error. The obtained results permit to address a classification of the error related to the used equipment. The paper ends by some suggestions proposed in order to minimize the errors.     

Study of optimal subset size in digital image correlation of speckle pattern images

This paper investigates the effect of subset size, associated with image pattern quality and subset displacement functions, on the accuracy of deformation measurements by digital image correlation(DIC). A concept of subset entropy is introduced in this work to quantify the subset image pattern quality for DIC analysis and its effectiveness was demonstrated by experimental studies. By employing white-light images with almost uniform subset entropy and first-order displacement functions, the effect of subset size on DIC analysis was investigated for the deformation cases of translation, uniform deformation, and simulated quadratic deformation, respectively. The results show that the chosen subset size must be large enough for precise displacement measurements when subset displacement functions match underlying actual deformation. On the other hand, optimal subset size in DIC for nonhomgeneous deformation measurements appears as a result of a tradeoff between the influence of random errors and systematic errors.     

Comparison of subset-based local and FE-based global digital image correlation: Theoretical error analysis and validation

Subset-based local and finite-element-based (FE-based) global digital image correlation (DIC) approaches are the two primary image matching algorithms widely used for full-field displacement mapping. Very recently, the performances of these different DIC approaches have been experimentally investigated using numerical and real-world experimental tests. The results have shown that in typical cases, where the subset (element) size is no less than a few pixels and the local deformation within a subset (element) can be well approximated by the adopted shape functions, the subset-based local DIC outperforms FE-based global DIC approaches because the former provides slightly smaller root-mean-square errors and offers much higher computation efficiency. Here we investigate the theoretical origin and lay a solid theoretical basis for the previous comparison. We assume that systematic errors due to imperfect intensity interpolation and undermatched shape functions are negligibly small, and perform a theoretical analysis of the random errors or standard deviation (SD) errors in the displacements measured by two local DIC approaches (i.e., a subset-based local DIC and an element-based local DIC) and two FE-based global DIC approaches (i.e., Q4-DIC and Q8-DIC). The equations that govern the random errors in the displacements measured by these local and global DIC approaches are theoretically derived. The correctness of the theoretically predicted SD errors is validated through numerical translation tests under various noise levels. We demonstrate that the SD errors induced by the Q4-element-based local DIC, the global Q4-DIC and the global Q8-DIC are 4, 1.8–2.2 and 1.2–1.6 times greater, respectively, than that associated with the subset-based local DIC, which is consistent with our conclusions from previous work.     

Strain field estimation based on digital image correlation and radial basis function

Two methods based on digital image correlation (DIC) and radial basis function (RBF) were proposed to obtain the accurate strain field in this paper. One is a combined method. RBF was applied to remove the noisy discrete displacement data first. After that, the strain was computed by a local least-squares algorithm. The other is a partial derivative of RBF (PD-RBF) based strain estimation method which integrated denoising with differential process. The effectiveness and accuracy of the proposed methods were verified through two numerical simulation experiments. A practical application on the normal strain measurement of an aluminum alloy beam under symmetric four-point bending via an outer loading frame was also presented. The measurement results are in good accordance with the data obtained by strain gauges. Furthermore, a shape parameter selection method based on rate of convergence was suggested. The new method simplifies the choice of the good shape parameter.     

Full-field digital image correlation with Kriging regression

A full-field Digital Image Correlation (DIC) method with integrated Kriging regression is presented in this article. The displacement field is formulated as a best linear unbiased model that includes the correlations between all the locations in the Region of Interest (RoI). A global error factor is employed to extend conventional Kriging interpolation to quantify displacement errors of the control points. An updating strategy for the self-adaptive control grid is developed on the basis of the Mean Squared Error (MSE) determined from the Kriging model. Kriging DIC is shown to outperform several other full-field DIC methods when using open-access experimental data. Numerical examples are used to demonstrate the robustness of Kriging DIC to different choices of initial control points and to speckle pattern variability. Finally Kriging DIC is tested on an experimental example.     

A weighting window applied to the digital image correlation method

In the digital image correlation method, two steps are used to calculate the displacements of tested images. The first step is to locate the integer-pixel displacement, and the next one is to compute the sub-pixel displacement based on the first step. The benefit from the computation method for the integer-pixel displacement is that its results directly affect the displacement accuracy of each point in tested images. In this paper, a new method is developed to calculate the displacement fields between two images taken before and after deformation of an object. A new correlation function with a weighting factor related to the position of each point in the image is taken into account. The analysis and experimental results confirm the validity of the new method. In addition, the calculation results from our method have more accuracy than that of the traditional one.     

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

The effect of out-of-plane motion (including out-of-plane translation and rotation) on two-dimensional (2D) and three-dimensional (3D) digital image correlation measurements is demonstrated using basic theoretical pinhole image equations and experimentally through synchronized, multi-system measurements. Full-field results obtained during rigid body, out-of-plane motion using a single-camera vision system with (a-1) a standard f55mm Nikon lens and (a-2) a single Schneider–Kreuznach Xenoplan telecentric lens are compared with data obtained using a two-camera stereovision system with standard f55mm Nikon lenses.Results confirm that the theoretical equations are in excellent agreement with experimental measurements. Specifically, results show that (a) a single-camera, 2D imaging system is sensitive to out-of-plane motion, with in-plane strain errors (a-1) due to out-of-plane translation being proportional to ΔZ/Z, where Z is the distance from the object to the pin hole and ΔZ the out-of-plane translation displacement, and (a-2) due to out-of-plane rotation are shown to be a function of both rotation angle and the image distance Z; (b) the telecentric lens has an effective object distance, Z_eff, that is 50× larger than the 55 mm standard lens, with a corresponding reduction in strain errors from 1250 μs/mm of out-of-plane motion to 25 μs/mm; and (c) a stereovision system measures all components of displacement without introducing measurable, full-field, strain errors, even though an object may undergo appreciable out-of-plane translation and rotation.     

Evaluation and improvement of digital image correlation uncertainty in dynamic conditions

Dealing with a moving target causes a motion effect (i.e. blurring) on the acquired images. This factor is an important source of measurement uncertainty while exploiting the Digital Image Correlation (DIC) technique. Therefore, the present study aims to evaluate and improve DIC uncertainty in dynamic conditions in the case of translating target. The study focuses on 2D DIC. In the case of 3D DIC similar problems arise, and therefore a complete understanding of two dimensional conditions will be of great help for further studies which deal with 3D conditions. The whole work can be divided into two main parts. In the first part, two different methods to simulate the motion effect on a reference image are proposed, discussed and validated. These methods allow simulating the acquired images in a real dynamic test and estimating the measurement uncertainty caused by the motion effect. The validation is performed by conducting several harmonic vibration tests with an electromagnetic shaker. In the second part of the study a numerical technique is proposed to estimate the motion effect present in an acquired image. This technique gives two main advantages. First of all, since the motion effect itself has a known influence on the uncertainty of measurement (first part of the study), we can predict the DIC׳s uncertainty by just having an acquired image. Furthermore, this numerical technique is used in the last part of this work to improve the performances of DIC in dynamic applications. In this way the bias error and the uncertainty of measurements were considerably decreased.     

Strain measurement of SU-8 photoresist by a digital image correlation method with a hybrid genetic algorithm

Digital image correlation (DIC) is a whole-field and non-contact strain measuring method. It could provide deformation information of a specimen by processing two digital images that are captured before and after the deformation. To search the deformed images, a hybrid genetic algorithm, in which a simulated annealing mutation process and adaptive mechanisms are added to the real-parameter genetic algorithm, is proposed in this work. To increase the accuracy and reliability of this method, some key parameters of this method are suggested. Then, this method is used to measure the strain during the micro tensile testing of SU-8 photoresist. In addition to the conventional single region, a double region is proposed to calculate the strain by DIC. The results indicate that while the strains obtained by single region are reasonable, those obtained by double region are accurate. Also the mechanical properties of SU-8 could be accurately obtained.     

Kalman filter based initial guess estimation for digital image correlation

In applications digital image correlation based algorithms often present a basis for analysis of movement/deformation of bodies. The sequence of the obtained images is analyzed for this purpose. Especially, in cases when the body׳s movement/deformation between two successive images is significant, the initial guess can have a major influence on the execution speed of the algorithm. In the worst case it can even cause the divergence of the algorithm. This was the inspiration to develop a new and unique approach for an accurate and reliable determination of an initial guess for each image pixel. Kalman filter has been used for this purpose. It uses past measurements of observed variable(s) for calculations. Beside that it also incorporates state space model of the actual system. This is one of the most important advantages provided by Kalman filter. The determined initial guess by the proposed method is actually close to the true one and it enables fast convergence. Even more important property of this approach is the fact that it is not path-dependant because each image pixel, which is defined in ROI, is tracked through the sequence of images based on its own past measurements and general state space model. Consequently, the proposed method can be used to analyze tasks where discontinuities between image pixels are present. The applied method can be used to predict an initial guess where reference and deformed subsets are related by translational and rotational motion. The advantages mentioned above are verified with numerical and real experiments. The experimental validations are performed by NR (Newton–Raphson) approach which is the most widely used. Beside NR method the presented algorithm is applicable for other registration methods as well. It is used as an addition for calculation of initial guesses in a sequence of deformed images.     

Incremental calculation for large deformation measurement using reliability-guided digital image correlation

Conventional digital image correlation (DIC) technique using a fixed reference image provides high-accuracy measurements but normally fails when serious decorrelation effect occurs in the deformed images due to large deformation, serious illumination fluctuations or other reasons. In this paper, an incremental reliability-guided digital image correlation (RG-DIC) technique, by combining the recently developed RG-DIC technique and an automatic reference image updating scheme, is proposed for large deformation measurement. In the incremental RG-DIC technique, a seed point is defined in the original reference image and searched in the deformed images, if the estimated correlation coefficient is larger than a preset threshold, which means no serious decorrelation effect exists in the deformed image, the RG-DIC technique is used to continue correlation analysis to obtain full-field displacements. Otherwise, the image recorded just before the current deformed image is chosen as an updated reference image to proceed with correlation analysis. Afterwards, the incremental displacements extracted by comparing the current deformed image and the updated reference image can be cumulated to determine the overall deformation. The effectiveness of the proposed technique is demonstrated by retrieving the full-field deformation of a foam sample subjected to large compressive deformation.     

Investigation of Portevin-Le Chatelier effect in 5456 Al-based alloy using digital image correlation

A variety of experimental methods have been proposed for Portevin-Le Chatelier (PLC) effect. They mainly focused on the in-plane deformation. In order to achieve the high-accuracy measurement, three-dimensional digital image correlation (3D-DIC) was employed in this work to investigate the PLC effect in 5456 Al-based alloy. The temporal and spatial evolutions of deformation in the full field of specimen surface were observed. The large deformation of localized necking was determined experimentally. The distributions of out-of-plane displacement over the loading procedure were also obtained. Furthermore, a comparison of measurement accuracy between two-dimensional digital image correlation (2D-DIC) and 3D-DIC was also performed. Due to the theoretical restriction, the measurement accuracy of 2D-DIC decreases with the increase of deformation. A maximum discrepancy of about 20% with 3D-DIC was observed in this work. Therefore, 3D-DIC is actually more essential for the high-accuracy investigation of PLC effect.     

高温下数字图像相关散斑最优成像探究

为了测量材料在高温甚至超高温下的力学性能,采用数字图像相关方法,并研究其在高温下的最优成像。采取不同的散斑制作方法,同时加入不同颜色的高温漆,在不同的温度节点,外加不同光源及相应的滤波片,采集并观察图像是否具有良好的对比度。普通的单色光源在800℃以后会逐渐失效,无法获取图像,而紫外光在1 200℃时依然可以获取较好的图像,且直接利用试件本身颜色作为底色效果更佳。采用紫外光照明可以实现DIC在高温环境下的测量。同时利用黑色或者蓝色散斑直接喷涂在试件上有着最佳的对比度,要优于常规的散斑制作方法。     

Speckle pattern quality assessment for digital image correlation

To perform digital image correlation (DIC), each image is divided into groups of pixels known as subsets or interrogation cells. Larger interrogation cells allow greater strain precision but reduce the spatial resolution of the data field. As such the spatial resolution and measurement precision of DIC are limited by the resolution of the image. In the paper the relationship between the size and density of speckles within a pattern is presented, identifying that the physical properties of a pattern have a large influence on the measurement precision which can be obtained. These physical properties are often overlooked by pattern assessment criteria which focus on the global image information content. To address this, a robust morphological methodology using edge detection is devised to evaluate the physical properties of different speckle patterns with image resolutions from 23 to 705 pixels/mm. Trends predicted from the pattern property analysis are assessed against simulated deformations identifying how small changes to the application method can result in large changes in measurement precision. An example of the methodology is included to demonstrate that the pattern properties derived from the analysis can be used to indicate pattern quality and hence minimise DIC measurement errors. Experiments are described that were conducted to validate the findings of morphological assessment and the error analysis.     

Image pre-filtering for measurement error reduction in digital image correlation

In digital image correlation, the sub-pixel intensity interpolation causes a systematic error in the measured displacements. The error increases toward high-frequency component of the speckle pattern. In practice, a captured image is usually corrupted by additive white noise. The noise introduces additional energy in the high frequencies and therefore raises the systematic error. Meanwhile, the noise also elevates the random error which increases with the noise power. In order to reduce the systematic error and the random error of the measurements, we apply a pre-filtering to the images prior to the correlation so that the high-frequency contents are suppressed. Two spatial-domain filters (binomial and Gaussian) and two frequency-domain filters (Butterworth and Wiener) are tested on speckle images undergoing both simulated and real-world translations. By evaluating the errors of the various combinations of speckle patterns, interpolators, noise levels, and filter configurations, we come to the following conclusions. All the four filters are able to reduce the systematic error. Meanwhile, the random error can also be reduced if the signal power is mainly distributed around DC. For high-frequency speckle patterns, the low-pass filters (binomial, Gaussian and Butterworth) slightly increase the random error and Butterworth filter produces the lowest random error among them. By using Wiener filter with over-estimated noise power, the random error can be reduced but the resultant systematic error is higher than that of low-pass filters. In general, Butterworth filter is recommended for error reduction due to its flexibility of passband selection and maximal preservation of the allowed frequencies. Binomial filter enables efficient implementation and thus becomes a good option if computational cost is a critical issue. While used together with pre-filtering, B-spline interpolator produces lower systematic error than bicubic interpolator and similar level of the random error. Cubic B-spline interpolator can achieve comparable efficiency as bicubic interpolator, while quintic B-spline interpolator requires about 1.5 times the running time.