基于立体视觉的结构变形全过程测量方法

发布时间：2018-06-19 00:14

本文选题：立体视觉 + 变形　；参考：《哈尔滨工业大学》2014年硕士论文

【摘要】：随着土木工程专业的高速发展，传统接触式测量手段在结构或构件的变形检测中逐渐暴露出效率低，测量范围有限等弊端。为了克服这些局限性，满足结构变形全过程测量要求，本文研究基于立体视觉的结构变形全过程测量方法。针对该方法，研究基于对角线长度的三步标定算法与，基于立体视觉的结构变形全过程测量算法。在上述算法基础上，编译立体视觉结构变形测量软件，，并进行试验研究。本文的研究内容主要包括： (1)对传统两步标定算法进行改进，提出基于对角线长度的立体视觉像机标定算法。本文在第二步中，采用LM算法对左右像机参数分别优化，以提高优化效率；在两步优化的基础上，以对角线长度作为约束条件，采用LM算法对像机外参数进行全局优化，缩放平移向量。编译了可以自动标定的三步优化标定软件。进行商业标定板对角线重建长度两步标定试验、商业标定板重建长度比较试验以及自制标定板重建长度比较试验，试验结果表明本文标定算法优于两步法。 (2)提出基于立体视觉的结构变形全过程测量算法。采用数字图像相关方法进行立体匹配和时序匹配，选择ZNCC作为相关函数以及曲面拟合的方法进行亚像素搜索；提出分块匹配的思想，减小多点匹配计算量；采用更新参考图的方法解决大变形情况下的测点误匹配问题；建立16节点的拉格朗日三次形函数，结合滑动平均的方法得到平滑的位移场；引入了格林应变张量，求解应变场。在上述算法基础上，编译立体视觉结构变形测量软件。 (3)完成304不锈钢试件拉伸试验和三层框架模型振动全程监测试验。将立体视觉方法测得的304不锈钢试件拉伸试验位移与应变数据，与引伸计及应变片的结果对比，验证本文算法的可靠性，在此基础上给出面外位移、位移场、应变场，验证算法的实用性；将立体视觉方法得到的三层框架模型振动全程监测数据与位移计结果比较，验证算法的可靠性，使用立体视觉方法对位移计撤掉或损坏以及结构倒塌情况下三层框架模型进行全程监测，验证本文算法用于结构变形全过程测量的可行性。
[Abstract]:With the rapid development of civil engineering specialty, the traditional contact measuring method has gradually exposed the disadvantages of low efficiency and limited measurement range in the deformation detection of structure or component. In order to overcome these limitations and meet the requirements of the whole process of structural deformation measurement, this paper studies the whole process measurement method based on the vertical body vision. In this method, the three step calibration algorithm based on diagonal length and the whole process measurement algorithm based on stereoscopic vision are studied. Based on the above algorithm, the software of the stereoscopic visual structure deformation measurement is compiled, and the experimental research is carried out.
The main contents of this paper are as follows:
(1) to improve the traditional two step calibration algorithm, a stereoscopic image machine calibration algorithm based on diagonal length is proposed. In this paper, in the second step, the LM algorithm is used to optimize the parameters of the left and right image machines to improve the optimization efficiency. On the basis of the two step optimization, the diagonal length is used as the constraint condition, and the LM algorithm is used for the external parameters of the image machine. The three step optimization calibration software that can be automatically calibrated is compiled. The two step calibration test of the diagonal reconstruction length of the commercial calibration board, the comparison test of the commercial calibration plate reconstruction length and the comparison test of the self-made calibration plate reconstruction length are carried out. The results show that the calibration algorithm is superior to the two step method.
(2) a complete process measurement algorithm for structural deformation based on stereoscopic vision is proposed. The digital image correlation method is used for stereo matching and timing matching, and ZNCC is selected as the correlation function and surface fitting method for subpixel search, and the idea of block matching is proposed to reduce the computation of multi point matching; the method of updating reference graph is used to solve the problem. The problem of mismatching of measuring points in large deformation is solved. The Lagrange three shape function of the 16 node is established, and a smooth displacement field is obtained with the method of sliding average. The Green strain tensor is introduced to solve the strain field. On the basis of the above algorithm, the deformation measurement software of the stereo vision structure is compiled.
(3) complete the tensile test of 304 stainless steel specimens and the whole course monitoring test of three story frame model. The data of displacement and strain of tensile test of 304 stainless steel specimens obtained by stereoscopic vision method are compared with the results of extensometer and strain gauge, and the reliability of the algorithm is verified. On this basis, out of plane displacement, displacement field and strain field are verified. The practicability of the algorithm is made. The reliability of the algorithm is verified by comparing the whole range monitoring data of the three layer frame model with the displacement meter. The three-dimensional vision method is used to monitor the displaced or damage of the displacement meter and the three layer frame model under the condition of structural collapse. The feasibility of process measurement.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU317

【相似文献】