首页 | 本学科首页   官方微博 | 高级检索  
     检索      

梯度算子选择对基于梯度的亚像素位移算法的影响
引用本文:潘兵,续伯钦,李克景.梯度算子选择对基于梯度的亚像素位移算法的影响[J].光学技术,2005,31(1):26-31.
作者姓名:潘兵  续伯钦  李克景
作者单位:中国科学技术大学,中科院材料力学行为与设计重点实验室,合肥,230027
基金项目:国家自然科学基金资助项目(10232010)
摘    要:基于梯度的亚像素位移算法中不同的灰度梯度计算方法对计算结果的影响进行了研究,列出了几种常用灰度梯度计算方法,并用对相邻5像素点的离散灰度进行三次样条插值的方法推导出两种灰度梯度算子。然后对计算机模拟生成的散斑图像对用各种梯度算子计算得出的亚像素位移与预先设定的真实值做了比较,给出了相应的均值误差和均值相对误差。最后的真实试件的刚体平移和单向拉伸实验结果表明,Barron算子是所列的几种梯度算法中最为精确、稳定的梯度算子。

关 键 词:数字图像相关  梯度算法  亚像素  样条插值
文章编号:1002-1582(2005)01-0026-06
修稿时间:2004年1月4日

Performance of gradient operators in algorithm of gradient-based subpixel registration
PAN Bing,XU Bo-qin,LI Ke-jing.Performance of gradient operators in algorithm of gradient-based subpixel registration[J].Optical Technique,2005,31(1):26-31.
Authors:PAN Bing  XU Bo-qin  LI Ke-jing
Institution:PAN Bing,XU Bo-qin,LI Ke-jingBehavior and Design of Materials of CAS,Hefei230027,China)
Abstract:Surface displacement measurement is important in experimental solid mechanics. Algorithm of gradient-based subpixel registration as one of the DIC methods is introduced to measure subpixel displacement, because of its high efficiency and precision. But different gradient operators in this algorithm have different performances. First, several existed gradient operators are illustrated and two gradient operators are deduced through cubic spline interpolation of five neighboring pixels. Then computer-simulated speckle images are used to verify the sensitivities of all the six gradient operators. The calculated error and relative error are presented. The algorithm is also calibrated using rigid body translation and uniform uniaxial tension test of metal specimen to examine the precision and stability. The final results show that the algorithm of gradient-based subpixel registration using Barron gradient operator stably gets the highest precision and is a practicable subpixel registration method.
Keywords:digital image correlation  gradient algorithm  sub pixel  cubic spline interpolation
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号