基于无人机的桥梁表观病害快速采集与识别

Rapid Collection and Recognition of Bridge Surface Defects Based on Unmanned Aerial Vehicle

摘要:

随着桥梁服役年限与服役桥梁数量的增长，传统人工桥梁检测逐渐满足不了日益增加的桥梁安全运维的需求。结合地理信息数据与无人机技术，配合倾斜摄影、激光雷达技术采集桥梁上部结构及下部数据，通过3D点云技术进行三维全息建模，采用深度学习算法对桥梁表观病害进行分类识别提取，采用计算机视觉技术对桥梁病害尺寸信息进行识别，实现桥梁表观病害的快速采集与智能化化识别。使用该方法对上海市G1503绕城高速胜辛路桥开展了桥梁表观病害无人机快速采集、三维建模与病害识别，取得了良好的应用效果。

关键词:无人机;桥梁检测;深度学习;目标检测;病害识别

Abstract:

With the increasing service life and number of bridges in service, the traditional manual bridge inspections are gradually unable to meet the growing demands of bridge safety operation and maintenance. Combined with the geographic information data and the unmanned aerial vehicle (UAV) technology, the data of the upper structure and lower structure of the bridge are collected by utilizing the oblique photography and laser radar technique. The 3D holographic modeling is carried out through 3D point cloud technology. The deep learning algorithm is used to classify, identify and extract the bridge surface defects. The computer vision techniques are utilized to identify the dimensions of bridge defects to realize the rapid collection and intelligent recognition of bridge surface defects. This method is applied to carry out the UAV-based rapid collection, 3D modeling and defect recognition of bridge surface defects for Shengxin Road Bridge in Shanghai G1503 Expressway. The good application results are achieved.

Key words:unmanned aerial vehicle (UAV); bridge inspection; deep learning; object detection, defect recognition.