摘要
为实现对铁路绝缘子污秽度的在线检测,提出一种基于支持向量数据描述和图像信息的污秽度异常检测方法。以人工涂污实验获得的绝缘子图像为基础,通过最大类间方差法分割图像得到绝缘子的盘面区域,计算颜色及纹理空间的特征,并利用核主元分析方法对特征向量进行融合与降维,最后通过支持向量数据描述方法实现污秽度的异常检测。结果表明,该方法可有效降低绝缘子污秽度的异常检测过程中的漏警率和虚警率,满足实际工作需求。
In order to achieve on-line detection of railway insulator contamination,a new method based on improved support vector data description(SVDD)and image information has been proposed.To obtain the disk area of insulator,the maximum interclass variance method(Otsu)is used to segment the polluted insulator images,which were obtained by artificial pollution smearing experiment.Then color and texture features are extracted by statistical method and the kernel principal component analysis(KPCA)algorithm is used to fuse and reduce the dimensionality of feature vector.Finally,the SVDD method is used to detect the abnormal contamination degree of the railway insulator.The results show that the method can effectively reduce the leakage rate and false alarm rate in the process of the contamination anomaly detection and meet the actual work requirements.
作者
吴文海
孙磊
柯坚
张霆
WU Wenhai;SUN Lei;KE Jian;ZHANG Ting(School of Mechanical Engineering,Southwest Jiaotong University,Chengdu 610031,China)
出处
《铁道标准设计》
北大核心
2019年第8期140-144,共5页
Railway Standard Design
关键词
铁路绝缘子
污秽度检测
核主成分分析
支持向量数据描述
模糊区域
railway insulator
contamination detection
kernel principal component analysis
support vector data description
fuzzy region
