摘要
针对无人机在电力输电线路巡检过程中,因空巡视角受风吹晃动导致线路难以完整呈现在巡检视频中,以及线路直径较小、观察距离远等限制,造成了巡检影像中线路识别准确率低下的问题.针对这一问题,提出了一种基于减帧的改进Mask-RCNN方法,该方法以相邻帧作为对比模板寻找图像差异从而提高识别准确度,并且通过减少运算过程中的数据处理量以减少数据处理时间,提高运算效率.在以复杂背景下导线为目标的试验中,本文方法的识别准确率与CPU使用率综合性能明显优于其他方法.
During the inspection of UAV power transmission lines, the line is difficult to be completely displayed in the inspection video due to the wind and shake of the aerial inspection view, and the small diameter of the line and the long observation distance. The problem of low line recognition accuracy in the inspection image arises. In response to this problem, an improved Mask-RCNN method based on frame reduction is proposed. The method in this paper uses adjacent frames as a comparison template to find image differences to improve recognition accuracy, reduce the calculation process, reduce the data processing time and improve the computational efficiency. In the experiment with the wire as the target under the complex background, the comprehensive performance of the recognition accuracy and CPU usage of the method in this paper are significantly better than other methods.
作者
陈麒
梁锐城
CHEN Qi;LIANG Ruicheng(Department of Mechanical and Electrical Engineering,Shantou Vocational and Technical College,Shantou 515041,Guangdong,China;College of Engineering,Shantou University,Shantou 515063,Guangdong,China)
出处
《汕头大学学报（自然科学版）》
2022年第2期43-49,共7页
Journal of Shantou University：Natural Science Edition
