Research On Digital Image Correlation Sub-pixel Displacement Measurement Algorithm

There are a variety of structural engineering deformation measurement method.Digital image correlation method by the laser speckle technique evolved as a new non-contact optical measurement method,because of its light path is simple,non-contact measurement,audience measurement,measurement and low environmental requirements,has been in engineering structures deformation measurement has been applied.But displacement measurement technique is based on digital image correlation pixels are measured,but the whole pixel measurement methods can not meet the basic requirements of precision.Therefore,digital image correlation analysis of sub-pixel measurement method has become one of the important research.Currently in a single sub-pixel accuracy for efficiency as well as lack of reconciliation,the need for further research to improve in order to meet the requirements of engineering structural deformation measurements.First,the summary of existing engineering structural deformation limitations of traditional measurement methods,pointed out the convenience of digital image correlation method.Summarizes the research status and application of digital image correlation method.Introduces the basic principles of digital image correlation method,measurement systems,pixel-Measurement representation of displacement conversion calibration method,correlation coefficient,characterizing the displacement mode,also describes the cross search method,a three-step search,coarse-fine search method,genetic algorithm integer pixel search algorithm.For production,quality evaluation speckle pattern speckle patterns made presentation.Second,because the entire sub-pixel is a pixel interpolation,it introduces the most adjacent domain interpolation,bilinear interpolation,bicubic interpolation interpolation method currently used for three sub-pixel displacement measurement algorithms: surface fitting,based on gradient subpixel displacement measurement algorithm(referred gradient method)and Newton-Raphson method(Newton-Raphson method,referred to as NR method)were introduced.Including surface fitting binary quadratic fit,bivariate cubic fitting and Gaussian fitting method;gradient method gradient method non-iterative gradient method and iteration.Again,interpolation accuracy and efficiency over several interpolation value method study comparison found Closest bilinear interpolation precision and better computational efficiency;to calculate the size of the sub-region,and fitting surface fitting function from the study showed that: when selecting 41 × 41 pixels to calculate the size of the sub-region,logarithmic functions and binary cubic fit fitting distance of 0.6 pixels,calculation results can maintain high accuracy and efficiency.Then,a new algorithm-Gradient-surface fitting algorithms.Found that the use of displacement measurement surface fitting,prone to calculation results in an error of about a half-pixel fluctuations,so that the displacement field in this position jump is large.Adopt proposed amendments Surface Fitting correlation matrix and reduce calculation errors in half pixel.Measurement practice found to obtain structural deformation when the speckle pattern,there may be overexposed or underexposed,the duration of exposure unstable or testing before and after the loading process is too long cause non-uniform illumination light intensity change that is tested under load and obvious differences in Light Field technical failures or objective conditions,which led to non-uniform changes in the measured surface image contrast and blur the image quality deterioration and other changes proposed to calculate the optical flow equation linear gradient descent sub-pixel displacement,reduced structural deformation before and after images of gray gradient change on the calculation.Finally,the use of digital simulation and experimental verification of the actual advantages of the above method in accuracy,computational efficiency and anti-noise area.