基于机器视觉和深度学习的空间结构球节点位移测量方法

Displacement measurement method of sphere joints in space structures based on machine vision and deep learning

随着信号处理和计算机技术的发展,机器视觉在科学研究和工程实践中得到了广泛的应用,在土木工程领域也越来越多地用于结构的应变和位移测量.但由于空间结构形态复杂、尺度大、测量操作繁琐等原因,机器视觉在空间结构位移测量领域还存在标记点布设难度大、目标点匹配困难、图像处理过程复杂等问题.近年来,随着以深度学习为代表的人工智能技术的快速发展,图像自动分类识别问题已经取得了巨大的突破,这也为空间结构位移测量提供了新的思路.本文提出一种基于深度学习的球节点自动检测和定位方法,并利用机器视觉原理对空间结构球节点进行图像匹配和三维重建,实现节点位移的测量.此方法可以快速对图像中所有球节点自动识别和定位,在保证精度的前提下大大减少图像处理时的工作量.通过对空间网架的球节点进行位移测量试验,验证了本方法可以完成空间结构球节点的三维位移测量任务,提高了测量效率,在空间结构位移监测领域有良好的应用前景.

As the development of signal processing technology and computer science, machine vision has been widely used in scientific research and engineering practice, especially in structural strain and displacement measurement in civil engineering. However, due to the complexity and large scale of space structures, using machine vision in displacement measurement of space structures has also raised some problems, such as labelling, target point matching and complicated image processing. Recently, with the rapid development of artificial intelligence technology represented by deep learning, there has been a great breakthrough in the field of automatic image classification and recognition, which provides a new method for displacement measurement. In this paper, a displacement measurement method for sphere joint using machine vision and deep learning algorithm is proposed, which can identify and locate all sphere joints in the image quickly and automatically, reducing a lot of workload without losing accuracy. A displacement measurement test of space truss was carried out to validate the accuracy and efficiency of the method. The results show that the measuring system is capable of performing the tasks of displacement measurement and has wide application prospects in displacement monitoring of space structures.