针对传统频域数字散斑相关法在测量物体面内位移测量精度较低的问题, 提出一种基于分形插值的频域数字散斑相关法。该方法在频域数字散斑相关法的基础上, 引入Hanning窗函数对图像进行滤波处理, 克服了图像边缘效应对最终位移值的影响; 同时利用散斑图像的子区域与整幅图像在结构形态和灰度特征的自相似性, 采用分形插值的方法进行亚像素插值, 进而能精确定位相关点, 得到更精确的亚像素位移值。实验结果表明: 该方法保持了频域散斑相关法的测量速度, 且将测量的绝对误差降低在0.01~0.03 pixel以内, 并进一步通过刚体平移实验测试了算法的可靠性。

For the low precision of in-plane displacement measurement using traditional digital speckle correlation methodin frequency domain, a new frequency domain correlation method based on fractal interpolation was proposed. Based on traditional digital speckle correlation in frequency domain, this method utilizes Hanning window function to filter the images in order to overcome the influence of edge effect on final displacement values. Considering theself-similarity of the sub-region and the whole imagein structure morphology and gray characteristics, fractal interpolation was adopted to improve the sub-pixel interpolation, which couldlocate relevant pointsand get more accurate sub-pixel displacement values. Experimental results show that this method could reduce the absolute error within 0.01 to 0.03 pixel and maintain the measurement speed as well. Moreover, the reliability of the algorithm is tested through the rigid translation experiments.