摘要
提出了一种新的考虑旋转位移分量的数字图像相关法——旋转相关匹配法,其特点是先对图像子区进行预旋转,再进行相关匹配、相关系数、整像素和亚像素位移的计算。在物体产生旋转位移分量时,该方法可有效提高变形前后图像子区间的相关系数和位移计算精度。通过斜拉伸、剪切和悬臂梁弯曲实验,对旋转相关匹配法和直接相关匹配法进行了对比研究,并用灰度梯度法计算旋转后新图像子区和变形后图像子区间的亚像素位移。结果表明,在旋转位移分量小于10°的情况下,相比于直接相关匹配法,旋转相关匹配法可以得到更高的相关系数;在用灰度梯度法计算亚像素位移条件下,位移精度与试件位移不产生旋转时的直接相关匹配相当。
A novel digital image correlation method taking into account rotating displacement component--spin correlation matching algorithm is proposed in this paper. Its characteristic is that image sub-region pre-rotation was performed firstly, and then the correlation matching, correlation coefficient and integer pixel and sub-pixel displacement were calculated. This method can effectively improve correlation coefficient and displacement calculation accuracy of image sub-regions before and after deformation, when specimen produces rotating displacement component. By oblique stretching, shearing and cantilever bending test, a comparative study between rotating correlation matching method and direct correlation matching method was carried out, and by gray-level gradient method, the sub-pixel displacement of new image sub-region after rotating and distorted image sub-region was calculated, respectively. Results show that when rotating angle is less than 10 degrees, spin correlation matching method can produces correlation coefficient with higher precision than that from direct correlation matching method. When gray-level gradient method was adopted to calculate sub- pixel displacement, the displacement precision obtained from spin correlation matching method is equal to that from direct correlation matching method, when specimen displacement does not produce rotation.
出处
《实验力学》
CSCD
北大核心
2013年第6期699-708,共10页
Journal of Experimental Mechanics
基金
国家重点基础研究发展计划资助项目(2012CB937500)
国家自然科学基金项目(10772171
11072235)
关键词
数字图像相关
旋转相关匹配
相关系数
灰度梯度
digital image correlation spin correlation matching correlation coefficient gray-level gra-dient
