首页|电力行业无人机巡检可见光图像与激光点云数据配准方法研究

电力行业无人机巡检可见光图像与激光点云数据配准方法研究

扫码查看
当前,电力行业为了提高输变电专业日常巡检的密度和精度,同时大幅降低输变电设备运维的人力成本,无人机技术被广泛引入到日常巡检业务中,通过无人机机载可见光成像设备和激光雷达设备获得了大量输变电设备可见光图像和激光点云数据.针对二维可见光图像数据深度信息丢失和三维点云数据智能识别障碍的问题,本文提出了一种可见光图像数据与激光点云数据的配准方法,即通过人工选取部分特征点的三维点云和二维像素点对数据,通过奇异值分解(SVD)方法求解出了关联三维世界坐标与二维像素坐标的参数矩阵T,验证结果表明经过上述矩阵T投影变换的二维特征点与其对应三维特征点吻合度较高,具备较好的配准精确度,基于此配准关系可实现输变电巡检点云数据与可见光数据的融合应用.
Research on the Registration Method of Visible Light Images and Laser Point Cloud Data for Unmanned Aerial Vehicle Inspection in the Power Industry
Currently,in order to improve the density and accuracy of daily inspections in the power transmission and transformation technology area,and significantly reduce the labor costs of operation and maintenance of power equipment,drone technology has been widely introduced into the inspection business in the power industry.And a large amount of visible light images and laser point cloud data of power transmission and transformation equipment have been obtained from drone onboard visible light imaging equipment and laser radar equipment yet.This paper proposes a registration method between visible light image data and laser point cloud data to address the issues of loss of depth information in two-dimensional visible light image data and obstacles in intelligent recognition of three-dimensional point cloud data.By manually selecting three-dimensional point clouds and data of two-dimensional pixel points for some feature point pairs,the parameter matrix T that associates the three-dimensional world with two-dimensional image points is solved using singular value decomposition(SVD)method.And the verification results show that the two-dimensional feature points transformed by the matrix T projection have a high degree of agreement with their corresponding three-dimensional feature points,which means a good registration accuracy.Based on this registration relationship,the fusion application of transmission and transformation inspection point cloud data and visible light data can be achieved.

Drone inspectionComputer visionVisible light imagesLaser point cloudData Registration

张少杰、赵李强、周静波、陈国坤、焦宗寒、杨伟、王欣、刘荣海

展开 >

云南电网有限责任公司电力科学研究院,云南 昆明 650032

云南电网有限责任公司楚雄供电局,云南 楚雄 675000

云南电网有限责任公司,云南 昆明 650011

无人机巡检 计算机视觉 可见光图像 激光点云 数据配准

云南电网有限责任公司科技项目

YNKJXM20220187

2024

云南电力技术
云南省电机工程学会 云南电力试验研究院(集团)有限公司电力研究院

云南电力技术

影响因子:0.244
ISSN:1006-7345
年,卷(期):2024.52(2)
  • 19