地铁隧道图像采集系统集成与病害检测技术

Image Acquisition System Integration and Disease Detection Technology for Subway Tunnel

首先梳理了地铁隧道病害检测现状,分析了国内外同类检测系统的优缺点,进而提出一种轻量化、模块化的地铁隧道表观病害检测系统。该系统集成线阵工业相机和智能运动平台,采用单个旋转相机代替传统阵列式相机,系统整体质量小于60 kg,可全自动采集地铁隧道表面图像,数据处理软件采用图像增强、加速稳健特征算法、加速区域生成卷积神经网络算法等算法,实现了全自动图像拼接和半自动病害提取。经在工程现场应用,系统检测速度达到2.5 km/h,裂纹检测精度达到0.2 mm,可全面检测地铁隧道表面裂纹、渗漏水等病害。

Firstly,the current situation of disease detection for subway tunnels was sorted out,the advantages and disadvantages of similar detection systems at home and abroad were analyzed,and then a lightweight and modular apparent disease detection system for subway tunnels was proposed.This system integrates line array industrial camera and intelligent motion platform,replacing the traditional array cameras with a single rotating one.The overall mass of the system is less than 60 kg.It can fully automate the acquisition of subway tunnel surface images,and its data processing software adopts image enhancement,speeded up robust features algorithms,faster region proposal convolutional neural network algorithms and other algorithms to realize automatic image stitching and semi-automatic disease extraction.After application in the engineering field,the detection efficiency of the system reaches 2.5 km/h,and the detection accuracy of crack reaches 0.2 mm,which can comprehensively detect the surface cracks and water leakage of subway tunnels.