数字体图像相关方法中基于迭代最小二乘法的物体内部变形测量  

Digital Volume Image Correlation by Using Iterative Least-Squares for Internal Deformation Measurement of an Object

作　　者：潘兵[1] 吴大方[1] 郭保桥[2] 

机构地区：[1]北京航空航天大学固体力学所,北京100191 [2]北京理工大学机电学院爆炸科学与技术国家重点实验室,北京100081

出　　处：《实验力学》2011年第6期665-673,共9页Journal of Experimental Mechanics

基　　金：国家自然科学基金(11002012;11172026);教育部高等学校博士学科点专项科研基金(20101102120015);教育部留学回国人员科研启动基金;中国航天科技集团公司航天科技创新基金(CASC201101);航空科学基金(2011ZD51043);中央高校基本科研业务费专项资金资助的课题

摘　　要：本文提出一种基于迭代最小二乘法的亚体素位移测量算法。该算法模型结合了符合实际情况的线性灰度变化模型和线性位移映射函数,采用仅需变形后体图像一阶灰度梯度的迭代最小二乘法计算亚体素位移和位移梯度。由于该算法采用的模型符合实际情况,因而具有更高的位移测量精度和更广泛的适用性。此外,由于仅需变形后体图像的一阶灰度梯度,无需计算其二级灰度梯度,因此极大地减小了计算复杂性。计算机模拟实验显示基于迭代最小二乘的亚体素位移测量算法原理简单、易于实现,具有较强的抗干扰能力,可用于真实体图像内部三维变形的高精度分析。A sub-voxel displacement measurement method based on iterative least-squares algorithm （ILS）, which combines practical gray scale linear-intensity-change model and linear displacement mapping function, is proposed in this paper to retrieve 3-D internal displacement vector field of an object. In this method, only the first -order gray-scale gradient of volume image after deformation is needed for ILS to measure the sub-voxel displacement and displacement gradient. This algorithm is capable of taking into account of both possible intensity changes and relative geometric transform of the target sub-volme, so it provides highest sub-voxel registration accuracy and widest applicability. Furthermore, this ILS algorithm significantly reduces computational complexity since only first -order gray-scale gradient of volume image after deformation is used. Computer-simulated volume images with controlled displacements were employed to investigate the performance of the proposed DVC method in terms of mean bias error and standard deviation error. Results reveal that the abovementioned method has rather strong anti-jamming capability and is capable of providing accurate measurement of 3-D internal deformation calculation in an easy-to-implement manner.

关 键 词：数字体图像相关方法 亚体素 内部变形测量 

分 类 号：O34[理学—固体力学]