Digital Image Correlation Method And Its Application On Mechanical Properties Measurement Of Materials

In modern optical metrology, digital image correlation (DIC) has become aneasy-to-use, practical and effective technique for quantitative full-field displacement andstrain measurement. The method has been successfully used in experimental mechanicsand other scientific fields due to the advantages of simple experiment facilities, strongenvironmental adaptability, wider measuring range, high automation and so on.Unfortunately, there are still some shortages exist on the integer pixel displacementsearching speed and the sub-pixel displacement computing accurate of DIC. Therefore, itis necessary to further investigate and improve the DIC method to solve complicatedengineering problems.A new integer pixel displacement searching algorithm of DIC was proposed, whichcombined by invariant moments and particle swarm optimization algorithm. Thisalgorithm overcame the shortages of the computational complexity of conventionalcorrelation function based on intensity and burden involved in point-to-point searchingmethod. The correlation function was established by six invariant moments of eachspeckle image in this algrithm. Correlating time was distinctly reduced due to the factthat six parameters represented the undeformed or deformed speckle image were used incorrelation function. The particle swarm optimization algorithm was used to optimizethis correlation function based on invariant moments. Therefore, correlating time wasfurther reduced. The validity and feasibility of the proposed algorithm were verified bysimulation experiments. The proposed algorithm, coarse-fine algorithm and geneticalgorithm were compared by simulation experiments in this dissertation. The resultsrevealed that the proposed algorithm had higher computational efficiency.In reality, the digital images are sometimes acquired with illumination lightingvariations, the instability offset of the imaging device, contrast variations during loadingand so on. Obviously, in these cases, the gray level intensity of physical points in theimages before and after deformation was changed. Consequently, an improved iterativegradient-based algorithm based on non-linear optical flow equation was proposed. Thenon-linear optical flow equation was employed in the algorithm with consideration of thenon-linear intensity variations of the deformed image, followed by an iterative leastsquares algorithm for calculating displacement field with sub-pixel accuracy. The results from numical experiments had shown that, among the most commly used iterativegradient-based algorithms, the improved algorithm displayed the most robust forenvironment variations and the highest computational efficiency. Then, two realtranslation experiments were conducted to compare the performance of the proposedalgorithm, curve-fitting method non-iterative gradient-based algorithm and N-Ralgorithm. The results had shown that the improved algorithm was insensitive toenvironment variations, and had the highest computing accuracy.In the image correlation process, the subset size was found to be critical to theaccuracy of the displacement measurement. In order to further improve the computingaccuracy of sub-pixel displacement measurement, a novel algorithm based on Shannonentropy for optimal selection of subset size. Six speckle patterns with different entropieswere numerically translated, and the displcements measured with DIC were comparedexact ones. Both mean errors and standard deviation of the measured displacements wereclosely related to Shannon entropy of speckle patterns used, and a so-called good specklepattern should be of large Shannon entropy. Consequently, Shannon entropy is a novelparameter for speckle pattern quality assessment, and can be used as guidance forpractical sample surface preparation. An optimal selection algorithm of subset size wasestablished by taking subset entropy and speckle pattern entropy as the indicator andthreshold, respectively. The numerical translation experiments were utilized to validatethe effectiveness of proposed algorithm for subset size selection. The resultsdemonstrated that computing accuracy of sub-pixel displacement was improvedeffectively by using the proposed algorithm and the subset mean size was reduced.The mechanical properties measurement system of materimals was developed basedon tracking speckles of sample surface by improved DIC method. Two CCD cameraswere employed to optically replace levers of clip gages in the measurement system. Totest the applicability, the mechanical stability and the accuracy of the system, the tensileexperiments with standard samples were performed. The results of tests indicated that thetheory was valid, the algorithm was feasible and high precision. It could completemechanical properties measurement of various materials.