基于空域形态一致性特征的绝缘子串定位方法

发布时间：2018-01-25 10:50

本文关键词： 绝缘子串目标定位显著性检测空域形态一致性特征描述子　出处：《中国电机工程学报》2017年05期 　论文类型：期刊论文

【摘要】：绝缘子串的精确定位是实现其故障诊断的重要前提。针对目前电网巡检图像中绝缘子串检测算法定位不精确的问题,该文提出了一种基于空域形态一致性特征的绝缘子串精确定位方法。首先,对巡检图像进行基于颜色对比和结构对比的显著性检测,以此确定绝缘子串候选区域;然后,对候选区域进行二值分割,分别沿水平方向和垂直方向进行二值像素累积投影,将图像矩阵转换为投影曲线;之后,从投影曲线中定义9个表征绝缘子串空域形态一致性特征的描述子;最后,以此描述子对候选区域进行绝缘子串搜索,实现绝缘子串精确定位。实验中,绝缘子串检测率达到92%,定位精确度达到0.93。结果表明,所提算法具有较好的识别正确性和定位精确性。
[Abstract]:Precise positioning of the insulator string is the important premise of fault diagnosis. According to the current insulator in power grid patrol images on the problem of inaccurate positioning detection algorithm, this paper proposes a series of precise positioning insulator consistency method based on the features of spatial form. First of all, the inspection image saliency detection contrast color contrast and structure based on, in order to determine the candidate region of insulator string; then, the candidate region of two value segmentation, respectively along the horizontal direction and the vertical direction of the two pixel value of the image matrix projection, projection curve; after the definition of spatial uniformity in morphology characteristics of 9 characterization of insulator string descriptors from the projection curve; finally this candidate region descriptor, search string, string to achieve precise positioning of insulator. In the experiment, the detection rate of 92% string, positioning accuracy reached 0. 93. results show that the proposed algorithm has good recognition accuracy and location accuracy.

【作者单位】：华北电力大学控制与计算机工程学院;
【基金】：国家自然科学基金项目(61401154) 中央高校基本科研业务费专项资金(2014MS140)~~
【分类号】：TP391.41;TM75
【正文快照】： 正确性和定位精确性。0引言随着智能电网和电力系统自动化的发展,计算机视觉技术越来越多地应用在电力设备的智能巡检和在线监测中,目标设备的识别和定位成为技术的关键[1-5]。其中,绝缘子是输电线路中的重要元件,起到电气绝缘及线路支撑的作用;同时也是故障频发元件,自爆、破

【相似文献】