中国电力2024,Vol.57Issue(9) :217-223.DOI:10.11930/j.issn.1004-9649.202306018

基于关键帧提取与图像识别的电网设施数字化重建

3D Reconstruction of Power Grid Facilities Based on Keyframe Extraction and Image Recognition

陈志永 韩璟琳 刘洋 李光毅 蒋雨 蔡瑞天
中国电力2024,Vol.57Issue(9) :217-223.DOI:10.11930/j.issn.1004-9649.202306018

基于关键帧提取与图像识别的电网设施数字化重建

3D Reconstruction of Power Grid Facilities Based on Keyframe Extraction and Image Recognition

陈志永 1韩璟琳 2刘洋 1李光毅 1蒋雨 3蔡瑞天4
扫码查看

作者信息

  • 1. 国网河北省电力有限公司经济技术研究院,河北石家庄 050021
  • 2. 国网河北省电力有限公司经济技术研究院,河北石家庄 050021;国网河北省电力有限公司,河北石家庄 050000
  • 3. 国网河北省电力有限公司雄安新区供电公司,河北保定 071000
  • 4. 华北电力大学电气与电子工程学院,北京 102206
  • 折叠

摘要

提出一种基于视频关键帧提取与图像识别的电网设施数字化重建方案.首先,通过移动作业终端拍摄待重建设施的视频序列,提取视频的关键帧并利用YOLOv5 网络识别并提取电力设施锚框去除环境背景,构造图像集;然后,以运动恢复结构为原理进行三维重建,提出改进定向快速旋转算法提高相机位姿计算精度,进而实现电网设施三维模型重建;最后,以变压器为例,验证所提方法可以有效完成电力设施三维模型的重建.

Abstract

A digital reconstruction scheme for power grid facilities based on video keyframe extraction and image recognition is proposed.Shoot a video sequence of the facility to be reconstructed through a mobile homework terminal,extract key frames from the video,and use YOLOv5 network to recognize and extract anchor frames of power facilities to remove environmental background,constructing an image set;Then,based on the principle of motion recovery structure,3D reconstruction is carried out,and an improved directional fast rotation algorithm is proposed to improve the accuracy of camera pose calculation.Thereby achieving the reconstruction of 3D models of power grid facilities;Finally,taking common transformers in the power grid as an example,it is verified that the proposed method can effectively complete the 3D model reconstruction of power facilities.

关键词

数字孪生/关键帧提取/图像识别

Key words

digital twins/keyframe extraction/image recognition

引用本文复制引用

基金项目

国家电网有限公司科技项目(5100-202113564A-0-5-SF)

出版年

2024
中国电力
国网能源研究院 中国电机工程学会

中国电力

CSTPCDCSCD北大核心
影响因子:1.463
ISSN:1004-9649
参考文献量25
段落导航相关论文