摘要
配电柜是动车组电气设备的关键设备之一,其中线缆的接线状态对电气设备的运行状态具有重大影响;传统的人工检测效率低,主观性强,配电柜背景复杂,光照条件恶劣,接线状态多样的特点,对线号识别造成较大的困难;针对配电柜的复杂情况提出了线号定位和分割的方法,首先利用基于尺度空间特征的SIFT算法实现线号定位,其次利用全局阈值和局部阈值进行分割,并进行角度校正和断裂字符修补,最后使用支持向量机(SVM)分类器进行线号识别,利用基于遗传算法的SVM参数优化方法;实现了对配电柜接线状态的识别,漏接识别率99.5%,错接识别率97.5%。
Distribution cabinet is a key equipment of the China Railway High-speed （CRH）, and connection state has significant effects on the operation of the electrical equipment. Traditional artificial detection has low efficiency and strong subjectivity and distribution cabinet always with complex background, poor lighting conditions and diversity of connection state which makes line number recognition more diffi- cult. This paper presents a location and segmentation method of line number, first using SIFT algorithm based on scale space feature to lo- cate the number, then using global and local threshold to binarize the image, followed by angle correction and fracture character repair, finally using support vector machine （SVM） to recognize line number and optimize parameters using genetic algorithm. This paper realized the recognition of connection state in the distribution cabinet, and get 99.5% recognition rate for not connect and 97.5% for wrong connect.
出处
《计算机测量与控制》
2016年第10期267-270,281,共5页
Computer Measurement &Control
关键词
接线状态
图像定位
图像分割
connection state
image locationt image segmentation
SVM
