深度学习在接触网定位器缺陷检测中的应用

张毅

深度学习在接触网定位器缺陷检测中的应用

张毅

中国铁路沈阳局集团有限公司供电处,沈阳 110001

详细信息

中图分类号: U225.4:TP39
计量
- 文章访问数: 95
- HTML全文浏览量: 2
- PDF下载量: 74
出版历程
- 收稿日期: 2017-09-10
- 刊出日期: 2018-03-14

Deep learning applied to defect detection of contact line locator

ZHANG Yi

Power Supply Department, China Railway Shenyang Group Co. Ltd., Shenyang 110001, China

摘要

摘要: 高速动车接触网运营安全的需求使得接触网关键零部件的缺陷自动检测成为一份有意义的工作。针对接触网巡检图像的定位器缺陷检测问题,本文提出了一种基于图像深度表示和直线检测的目标检测一体化算法。该算法采用选择搜索算法获得定位器在图像中可能存在的备选区域,利用深度卷积神经网络计算图像的深度特征,通过多任务学习的算法求得定位器的局部区域。随后,利用Canny边缘提取和Hough直线检测的方法在局部区域内精确检测定位器直线。针对接触网巡检图像的实际应用场景,对该算法在不同场景下进行验证,试验结果表明,该算法可以有效解决实际场景下的定位器缺陷检测问题。
- 定位器检测 /
- 缺陷检测 /
- 深度学习 /
- 机器学习
Abstract: Aiming at the locator defect detection of contact line inspection image, this paper proposed a detection integration algorithm based on image depth representation and line detection. This algorithm adopted selective search algorithm to obtain the possible alternative regions in image where locator may exist. The deep features of image were calculated by deeply convolutional neural network and local region of locator could be obtained by multi-task learning algorithm. Subsequently, by using Canny edge extraction and Hough line detection method, locator line could be exactly detected in local region. According to the practical application scene of contact line inspection image, this algorithm has been tested by several locator detection tasks at different scenes. The experiments proved that this algorithm could solve the detector defect detections problems effectively in complex environment.
- locator detection /
- defect detection /
- deep learning /
- machine learning

HTML全文

参考文献(11)

[1]	罗健. 高速铁路接触网定位器坡度问题的深化研究[J],铁道工程学报,2013（1）：76-81.
[2]	Nasrabadi N M. A stereo vision technique using curve-segments and relaxation matching[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 1992(5): 566-572.
[3]	Lowe D G. Distinctive image features from scale-invariant keypoints[J]. International journal of computer vision, 2004, 60(2): 91-110.
[4]	Felzenszwalb P F, Girshick R B, McAllester D, et al. Object detection with discriminatively trained part-based models[J]. IEEE Transactions on Pattern Analysis and Machine Intel-ligence, 2010, 32(9): 1627-1645.
[5]	LeCun Y, Bottou L, Bengio Y, et al. Gradient-based learning applied to document recognition [J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.
[6]	Krizhevsky A, Sutskever I, Hinton G E. Imagenet classification with deep convolutional neural networks[C]//Advances in neural information processing systems. 2012: 1097-1105.
[7]	Sermanet P, Kavukcuoglu K, Chintala S, et al. Pedestrian detection with unsupervised multi-stage feature learning[C]// IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2013: 3626-3633.
[8]	Gu C, Lim J J, Arbeláez P, et al. Recognition using regions[C]// IEEE Conference on. IEEE on Computer Vision and Pattern Recognition, 2009. CVPR 2009: 1030-1037.
[9]	李文辉. 航拍视频中运动目标的检测与跟踪算法研究[D].西安：西安电子科技大学,2014.
[10]	王海地. 基于显著性检测和分类器训练的航拍图像车辆检测[D].北京：北京交通大学,2015.
[11]	Jia Y, Shelhamer E, Donahue J, et al. Caffe: Convolutional architecture for fast feature embedding[C]//Proceedings of the ACM International Conference on Multimedia. ACM, 2014: 675-678.

施引文献

资源附件(0)

计量

文章访问数: 95
HTML全文浏览量: 2
PDF下载量: 74
被引次数: 0

深度学习在接触网定位器缺陷检测中的应用

计量

出版历程

Deep learning applied to defect detection of contact line locator

计量

出版历程

目录