[1]杨 涛,李 祎,陈晶华,等.基于背景差分的巡检机器人视觉识别方法[J].机械与电子,2020,(12):60-64.
　YANG Tao,LI Yi,CHEN Jinghua,et al.Key Technologies of Inspection Robot Video System Based on Background Difference[J].Machinery & Electronics,2020,(12):60-64.
点击复制

基于背景差分的巡检机器人视觉识别方法()
分享到：

机械与电子[ISSN:1001-2257/CN:52-1052/TH]

卷:
期数:
2020年12期
页码:
60-64
栏目:
智能工程
出版日期:
2020-12-18

文章信息/Info

Title:
Key Technologies of Inspection Robot Video System Based on Background Difference
文章编号:
1001-2257（2020）12-0060-05
作者:
杨 涛1; 李 祎1; 陈晶华1; 文 炜2; 张香怡3; 顾继俊3

1. 中海石油(中国)有限公司，北京 100010；

2.中石化川气东送天然气管道有限公司，湖北 武汉 430020；

3.中国石油大学（北京）机械与储运工程学院，北京 102200

Author(s):
YANG Tao1; LI Yi1; CHEN Jinghua1; WEN Wei2; ZHANG Xiangyi 3; GU Jijun 3
1.CNOOC (China) Co., Ltd. Beijing 100010 , China;
2. Sinopec Sichuan-East Gas Pipeline Co., Ltd.,Wuhan 430020,China;
3. College of Mechanical and Storage Engineering, China University of Petroleum (Beijing),Beijing 102200 ,China
关键词:
巡检机器人; 视频对齐; 像素级匹配; 背景差分; 异物检测
Keywords:
inspection robot;  video alignment;  pixel-level matching;  background difference;  foreign object detection
分类号:
TP242.2
文献标志码:
A
摘要:
针对巡检机器人检测异物范围受限，只能检测经过训练的异常物体问题，提出了一种基于视频对齐的背景差分技术。巡检机器人由于运行速度不稳定、摄像头角度存在偏差，会导致视频无法对齐，识别准确率低，因此采用一种改进的联合优化算法对视频进行处理。首先使用改进的DTW算法逐帧查找与目标视频对应的帧，将视频对应帧提取出来；再采用SURF算法提取对应帧的特征点，进行筛选后使用DeepFlow算法得到对应帧图像的变形场矩阵，进而将图像角度修正，并进行像素级的匹配。实验结果表明，DTW-SURF-DeepFlow联合算法可以将图像进行完全的对齐，经过对齐的图像通过背景差分即可以检测出任意的异常物体，提高了异物检测的准确性。
Abstract:
Aiming at the problem that the inspection robot has limited detection range of foreign objects and can only detect abnormal objects after training, a new background difference technology based on video alignment is proposed. Due to the unstable running speed of the inspection robot and the deviation of the camera angle, the video cannot be aligned and the recognition accuracy is low. Therefore, an improved joint optimization algorithm is used to process the video. First, use the improved DTW algorithm to find the frame corresponding to the target video frame by frame, and extract the corresponding frame of the video; then use the SURF algorithm to extract the feature points of the corresponding frame, and after filtering, use the DeepFlow algorithm to obtain the deformed field matrix of the corresponding frame image, and then Correct the image angle and perform pixel-level matching. Experimental results show that the DTW-SURF- DeepFlow joint algorithm can completely align the images, and the aligned images can detect any abno rmal objects through the background difference, which improves the accuracy of foreign object detection.

参考文献/References:

[1]田蕴,李帅,王真.智能巡检机器人的发展与设计趋势探析[J].工业设计,2019(11):143-144.

[2]吕小文.悬挂轨道式智能巡检机器人研究[D].杭州：浙江大学,2019.

[3]张磊,彭富波,李嘉峻.基于施工现场动态视觉监控系统研究应用[J].云南水力发电,2019,35(6):177-181.

[4]李曼.基于机器学习的故障识别方法与系统研制[D].北京：北京交通大学,2015.

[5]MUKOJIMA H ,DEGUCHI D ,KAWANISHI Y,et al.Moving camera background-subtraction for obstacle detection on railway tracks [C]//2016 IEEE International Conference on Image Processing,2016:3967–3971.

[6]Freitas R S,Xaud M F S,Marcovistz I,et al. The embedded electronics and software of DORIS offshore robot[J].IFAC-PapersOnLine, 2015,48(6):208-213.

[7]曾富豪,刘润琴,董婷.CAN通信在BMS中的研究与应用[J].汽车实用技术,2019(12):49-51.

[8]郭义凡.移动摄像下目标检测与跟踪研究[D].厦门：厦门大学,2014.

[9]GALASSI M ,R?YR?Y A ,DE CARVALHO G P S,et al. DORIS：A mobile robot for inspection and monitoring of offshore facilities[C]// 2014 Anais do XX Congresso Brasileiro de Automática,2014：3174-3181.

[10]DA SILVA A F ,THOMAZ L A ,NETTO S L,et al.Online video-based sequence synchronization for moving camera object detection[C]//2017 IEEE 19th International Workshop on Multimedia Signal Processing,2017 :1-6.

[11]金斌英.基于改进的SURF图像配准算法研究[J].计算机测量与控制,2019,27(11):228-232.

[12]WEINZAEPFEL P ,REVAUD J ,HARCHAOUI Z,et al.DeepFlow: Large displacement optical ?ow with deep matching[C]// Proceedings of the 2013 IEEE International Conference on Computer Vision,2013:1385–1392.

相似文献/References:

[1]邹 林.基于 ＲＯＳ的电力变电站巡检移动机器人[J].机械与电子,2019,(05):76.
　Mobile Robot for Power Substation Patrol Inspection Based on ROS[J].Machinery & Electronics,2019,(12):76.
[2]王兴涛,李文永,陈 亮.用于电力巡检机器人的挂空异物智能检测方法[J].机械与电子,2021,(12):76.
　WANG Xingtao,LI Wenyong,CHEN Liang.Intelligent Detection Method of Hanging Foreign Matter for Power Inspection Robot[J].Machinery & Electronics,2021,(12):76.
[3]李 鹏,闵小翠,王建华.基于改进蚁群算法的巡检机器人避障路径规划方法设计[J].机械与电子,2022,(02):71.
　LI Peng,MIN Xiaocui,WANG Jianhua.Design of Obstacle Avoidance Path Planning Method for Inspection Robot Based on Improved Ant Colony Algorithm[J].Machinery & Electronics,2022,(12):71.
[4]王兴涛,唐永成,李文永.基于视觉伺服的巡检机器人返航路径确定方法[J].机械与电子,2022,(03):71.
　WANG Xingtao,TANG Yongcheng,LI Wenyong.Method for Determining Return Path of Inspection Robot Based on Visual Servoing[J].Machinery & Electronics,2022,(12):71.
[5]王艳阳,熊 威,谭 娟,等.基于巡检机器人的变电站设备运行状态监测技术[J].机械与电子,2022,(09):71.
　WANG Yanyang,XIONG Wei,TAN Juan,et al.Monitoring Technology of Substation Equipment Operation Status Based on Inspection Robot[J].Machinery & Electronics,2022,(12):71.
[6]朱李超,施 会,茅晓亮.基于动态特征点的巡检机器人目标定位系统设计[J].机械与电子,2022,(11):71.
　ZHU Lichao,SHI Hui,MAO Xiaoliang.Design of Target Location System for Inspection Robot Based on Dynamic Feature Points[J].Machinery & Electronics,2022,(12):71.
[7]朱敏捷,李建华,周灵刚.基于巡检机器人的电缆卷筒减速器故障诊断技术[J].机械与电子,2022,(12):20.
　ZHU Minjie,LI Jianhua,ZHOU Linggang.Fault Diagnosis Technology of Cable Reel Reducer Based on Inspection Robot[J].Machinery & Electronics,2022,(12):20.
[8]马宏伟,殷鸿鑫,范 帆.动车组底部智能巡检机器人机械结构参数优化[J].机械与电子,2022,(12):72.
　MA Hongwei,YIN Hongxin,FAN Fan.Optimization of Mechanical Structure Parameters of Intelligent Inspection Robot at the Bottom of Multiple Units[J].Machinery & Electronics,2022,(12):72.
[9]孟延辉,史晓龙,王占宁,等.全天候移动车间巡检机器人目标定位算法[J].机械与电子,2023,41(02):71.
　MENG Yanhui,SHI Xiaolong,WANG Zhanning,et al.All-weather Mobile Workshop Inspection Robot Target Localization Algorithm[J].Machinery & Electronics,2023,41(12):71.
[10]梁海宁.热电厂巡检机器人障碍物识别与避让控制方法[J].机械与电子,2023,41(06):36.
　LIANG Haining.Obstacle Recognition and Avoidance Control Method for Inspection Robot in Thermal Power Plant[J].Machinery & Electronics,2023,41(12):36.

备注/Memo

备注/Memo:
收稿日期：2020-08-25
基金项目：天然气管道智能视频监控系统研究-系统数据研发(35150573-20-ZC0607-0002)；国家重点研发计划项目（2016YFC0303706，2017YFC0805803）
作者简介：杨 涛(1978—)，男，河北保定人，高级工程师，研究方向为智能油气田、无人平台等。
更新日期/Last Update: 2020-12-18