大型结构变形实时摄像测量方法

本文对实时测量大型结构静力变形实验中如何应用摄像测量方法展开研究。针对实时摄像测量的两个主要难题研究解决方法,即图像特征点自动提取和多像机图像特征点自动对应。通过在待测结构表面粘贴LED发光灯或使用激光光栅投影做为合作标志点,提出NNLOG(normalnized negative laplacian of gaussian)算子用于图像上合作标志点的实时提取与跟踪;根据多摄像机约束关系,构造对极几何约束代价函数,解决多像机图像特征点之间的对应问题。设计实现了精度实验,表明位移测量误差小于1mm,相对误差小于1%,基于本文方法实现的测量系统已应用于某型飞机静力变形实验。

A Real-time Camera Measurement Method for Large-scale Structure Deformation

This paper studies the application of camera measurement method in large-scale structure real-time static deformation experiment. Study focuses on two main challenges from real-time camera measurement, namely, automatic extraction of image feature points and automatic correspondence of multi-camera image feature points. LED (light-emitting diode) lights were pasted on tested structure surface, or laser raster projection was used as cooperative mark points, then NNLOG (normalnized negative laplacian of gaussian) operator was developed for automatic extraction and tracking of cooperative mark points on images. An epipolar geometry constraint optimization method was proposed, and a cost function based on the epipolar geometry constraint was constructed to solve image point correspondence between multiple cameras. A precision experiment was implemented. Results show that the error of deformation measurement is less than 1 mm and relative error is less than 1%. A measurement system realized based on the method presented in this paper has already applied for static deformation test of an airplane.