架空输电线路无人机巡检图像缺陷识别方法研究  

Research on Defect Recognition Method for Unmanned Aerial Vehicle Inspection Images of Overhead Transmission Lines

作　　者：王涛[1] 严永锋 汪滢[1] 任涛[1] 田成 李文 Wang Tao;Yan Yongfeng;Wang Ying;Ren Tao;Tian Cheng;Li Wen(Wuhan Electric Power Vocational and Technical College,Wuhan,China)

机构地区：[1]武汉电力职业技术学院,湖北武汉

出　　处：《科学技术创新》2024年第24期132-135,共4页Scientific and Technological Innovation

摘　　要：传统人工缺陷图片判读存在效率低下、过度依赖个人经验、识别率低等问题,采用一种基于深度学习的目标检测算法对巡视照片进行模拟训练学习,实现对输电设备部件的识别定位,采用超分辨率算法(SRCNN)对具体部位缺陷进行自主识别,对提高无人机巡检数据照片处理效率具有重要意义。Traditional manual defect image interpretation has problems such as low efficiency,excessive reliance on personal experience,and low recognition rate.A deep learning based object detection algorithm is used to simulate and train inspection photos to achieve recognition and positioning of transmission equipment components.The super-resolution algorithm(SRCNN)is used to autonomously identify specific defects,which is of great significance for improving the efficiency of drone inspection data photo processing.

关 键 词：深度学习 超分辨率算法 自主识别 

分 类 号：TN99[电子电信—信号与信息处理]