摘要
针对目前大部分铁路扣件采用人工巡检的情况,提出基于二维Gabor变换和胶囊网络相结合的铁路扣件状态检测方法。在获得扣件原始图像的基础上采用二维Gabor变换进行滤波,分析滤波结果,通过胶囊网络进行状态识别,并对比分析滤波后产生的图像在卷积神经网络、胶囊网络以及两者结合的神经网络下所产生的不同检测结果,得出最优检测方法。研究表明,该方法可代替大部分人工检测,有效提高检测效率,保障行车与巡检人员安全。
Because most of the railway fasteners need to be manually inspected,this paper proposes an automatic detection method for the railway fasteners based on 2D Gabor transformation and capsule network.Based on the original image of the fastener,the 2D Gabor transform is used to filter and analyze the filtering results.Then,the state recognition is performed through the capsule network.It also compares and analyzes the different detection results produced by the image after filtering in the convolutional neural network,the capsule network and the neural network combined with them,which are used to obtain the best detection method.Research result shows that this method can be used for most of the manual detection.The detection efficiency is improved greatly and the safety of driving and inspection personnel is ensured.
作者
王笑冬
林建辉
WANG Xiaodong;LIN Jianhui(State Key Laboratory of Traction Power,Southwest Jiaotong University,Chengdu 610031,China)
出处
《机械制造与自动化》
2020年第6期134-137,共4页
Machine Building & Automation
关键词
二维Gabor变换
胶囊网络
卷积神经网络
铁路扣件
自动检测
2D Gabor transform
capsule network
convolutional neural network
railway fasteners
automatic detection
