基于机器视觉的隧道衬砌表观病害检测技术研究进展

Research Progress of Machinery Visual-based Tunnel LiningAppearent Disease Detection Technology

针对隧道衬砌表观裂缝病害的智能化检测算法问题,梳理了自动检测算法的处理流程,介绍了3种裂缝自动检测的方法:图像处理、传统机器学习、深度学习算法,并分析了3种方法的适用范围及优缺点,重点讨论了深度学习算法,它从根本上改变了裂纹检测的方式,大幅提高了检测性能。对基于分类、目标检测和分割的深度学习神经网络在裂纹检测中的应用进行了综述和比较,展望了该算法在裂缝自动检测的应用前景。

Aiming at the problem of intelligent detection algorithm of tunnel lining apparent crack disease,this paper summarizes the processing flow of automatic detection algorithm,and reviews three main methods of automatic crack detection and their scopes of application and merits and demerits:image processing,traditional machine learning and deep learning algorithm.The deep learning algorithm method is mainly introduced,which changes the way of crack detection and greatly improves the detection performance.The application of deep learning neural network based on classification,object detection and segmentation in crack detection is reviewed and compared,and its application of segmentation algorithm on automatic crack detection is prospected.