在数字图像相关方法对物体形变测量中,FA-NR算法实现了高精度测量,IC-GN算法在此基础上提高了测量效率.为进一步提升测量精度,提出了一种基于IC-GN的改进算法(GIC-GN).在已知整像素初始位置上,通过梯度法求得更准确的亚像素位移,减小Hessian矩阵计算过程中产生的误差,同时加快迭代的收敛速度,有效提高了测量精度和效率.仿真实验验证结果表明,GIC-GN算法误差能够稳定在10-4～10-3 pixel之间,对比IC-GN算法精度提升了 10％～60％,耗时是FA-NR算法的0.1倍、IC-GN算法的0.8倍,能够实现对物体形变信息的高精度、高效率测量.
In the measurement of object deformation using digital image correlation methods,the FA-NR algorithm achieves high-precision measurement,while the IC-GN algorithm improves measurement efficiency on this basis.To further improve measurement accuracy,an improved algorithm based on IC-GN(GIC-GN)was proposed.By using the gradient method to obtain a more accurate sub pixel displacement at the known initial position of the entire pixel,the error generated in the Hessian matrix calculation process is reduced,and the convergence speed of the iteration is accelerated,effectively improving measurement accuracy and efficiency.The simulation experiment verification results show that the error of GIC-GN algorithm can be stable between 10-4-10-3 pixel,and the accuracy of IC-GN algorithm is improved by 10％-60％compared to FA-NR algorithm.The time consumption is 0.1 times that of FA-NR algorithm and 0.8 times that of IC-GN algorithm,which can achieve high-precision and efficient measurement of object deformation information.
孙泽刚;张奎;蒋强;黎军华
四川轻化工大学机械工程学院,四川宜宾 644002乐山师范学院电子信息与人工智能学院,四川乐山 614000乐山一拉得电网自动化有限公司,四川乐山 614000
数字图像相关;IC-GN算法;Hessian矩阵;高精度;高效率
digital image correlation;IC-GN algorithm;Hessian matrix;high precision;high efficiency
《影像科学与光化学》 2024 (001)
9-15 / 7
四川省科技计划项目(2022YFG0075)
10.7517/issn.1674-0475.230804
评论