关键词: 数字图像相关方法, 测点位置, 优化, 平均灰度梯度, 散斑质量, 格林样条插值

Abstract: In traditional Digital Image Correlation（DIC）methods, points are regularly distributed, that is, points are parallel in horizontal and vertical directions. Thus, a few points can be distributed in positions where speckle qualities are bad. In this paper, a digital image correlation method based on the optimization of positions of points, Newton-Raphson（N-R）iteration and particle swarm optimization is proposed. Firstly, the positions of original points are optimized; secondly, the displacement field for the irregular optimal points is obtained by use of the N-R iteration and PSO; finally, the displacement field is interpolated using two-dimensional Green’s function-based interpolator to obtain the displacement field at regularity distributed points, and then, the strain field is obtained by use of the central difference method. For three speckle images where speckle qualities are greatly different, positions of points are optimized, and comparisons between the present and traditional results are conducted. It is found that the present method has reasonable accuracy for speckle images with the strain of 0.01~0.05, the size of subsets of 21~41 pixel, and the mean intensity gradient of 10~20 pixel-3, which is related to the point optimization. It is expected that high accurate results can be obtained using the present method if only the positions of points distributed at positions with bad speckle qualities are optimized.

Key words: digital image correlation method, positions of points, optimization, mean intensity gradient, speckle quality, Green’s function-based interpolator