首页 | 本学科首页   官方微博 | 高级检索  
     

融合目标检测的交互式电网设备图像分割算法
引用本文:叶健强,张晓晨,孙敦虎,李荆晖,张麒腾,张明璨. 融合目标检测的交互式电网设备图像分割算法[J]. 宁夏电力, 2022, 0(1): 19-24
作者姓名:叶健强  张晓晨  孙敦虎  李荆晖  张麒腾  张明璨
作者单位:国网宁夏电力有限公司建设分公司,宁夏银川750004;西安交通大学电气学院,陕西西安710000
基金项目:国网宁夏建设分公司2021年群众性科技创新项目(5229JS210001)
摘    要:图像处理技术在电网中有着十分重要的应用,图像分割是图像处理技术的重要组成部分.本文提出了一种融合目标检测的交互式电网设备图像分割算法,在传统的图像抓取(GrabCut)交互式分割算法的基础上,使用目标检测得到的检测标定框代替用户选择的初始化框,在目标检测算法上采用目前已知开源最快的最轻量的你只看一次(YOLO-Fast...

关 键 词:目标检测  交互式分割算法  电网设备  图像处理

An interactive grid equipment image segmentation algorithm integrating target detection
YE Jianqiang,ZHANG Xiaochen,SUN Dunhu,LI Jinghui,ZHANG Qiteng,ZHANG Mingcan. An interactive grid equipment image segmentation algorithm integrating target detection[J]. Ningxia Electric Power, 2022, 0(1): 19-24
Authors:YE Jianqiang  ZHANG Xiaochen  SUN Dunhu  LI Jinghui  ZHANG Qiteng  ZHANG Mingcan
Affiliation:Construetion Braneh of State Crid Ningxia Eleetrie Power Co.,Ltd.,Yinehuan Ningxia 750004 ,China;Sehoool of Eleetrieal Engineering,Xi'' an Jiaotong U niversity,Xi'' an Shanxi 710000 ,China
Abstract:Image processing technology has a very important application in power grid, and image segmentation is an important part of image processing technology. In this paper, an interactive power grid equipment image segmentation algorithm integrating target detection is proposed. Based on the traditional GrabCut interactive segmentation algorithm, the detection calibration box obtained by target detection is used to replace the initialization box selected by the used On the target detection algorithm, the fastest and lightest YOLO-Fastest algorithm known to be open source is adopted, which can overcome the difficulty of insufficient computing power of many equipment. The applicability of the algorithm is wider. Experimental results show that compared with the traditional interactive segmentation algorithm, this algorithm has shorter execution time, fewer interaction times, and has good stability.
Keywords:
本文献已被 万方数据 等数据库收录!
点击此处可从《宁夏电力》浏览原始摘要信息
点击此处可从《宁夏电力》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号