拉索机器人检测拉索PE表观图像数据

拉索机器人在斜拉索上爬行过程中，实时拍摄斜拉索表观图像。机器人控制系统在控制机器人爬行过程中，每间隔一定距离控制拉索周向4个工业相机同时对拉索PE外观进行一次拍摄，获取到一组4张图像，图像文件名以相机编号、离拉索底部的位移、时间命名，可确保后期通过图像文件的名字获取到图像文件的拍摄相机，拍摄位置离拉索底部的距离及拍摄时间，从而更方便地定位病害位置。

During the crawling process of a cable robot along a stayed cable, real-time capture of the surface images of the stayed cable is conducted. While controlling the robot's crawling movement, the robot control system triggers four industrial cameras arranged circumferentially around the stayed cable to simultaneously capture images of the cable's PE sheath surface at fixed distance intervals, yielding a set of 4 images each time. The filenames of these images are named with the camera ID, the distance from the bottom of the stayed cable at the shooting position, and the capture timestamp. This naming convention allows the capturing camera, the shooting position's distance from the stayed cable bottom, and the capture time to be retrieved from the filename during post-processing, thus facilitating the accurate localization of the stayed cable's defect locations.

搜集汇总

背景与挑战

背景概述

该数据集由香港中文大学（深圳）创建，包含拉索机器人在斜拉索上爬行过程中实时拍摄的PE表观图像，用于检测拉索病害。数据采集时每间隔一定距离使用4个工业相机同时拍摄，生成一组4张图像，文件名编码了相机、位置和时间信息，便于精确定位病害，总数据量为13.84GB，包含3773个文件，属于国家重点研发计划项目'大跨度桥梁检测作业机器人'的成果。