摘要
由于传统的驱鸟方法效果较差,通过研究鸟类检测,可以配合驱鸟设备实现高效的智能化驱鸟,提高驱鸟率。针对变电站的复杂环境,文中采用KNN背景差分法获取运动目标,并将运动目标进行分割和重组,提取图像中的感兴趣区域,最后利用优化的YOLOv4网络对感兴趣区域进行鸟类检测。实验结果表明,所提方法可以较为准确高效地检测变电站中运动的鸟类目标,具有较强的实用性。
Due to the poor performance of traditional bird driving methods,the use of bird driving equipment can realize the efficient intelligent bird driving after the study of bird detection.Based on the complex environment of the substation,this paper uses KNN background difference method to extract moving targets,and then divides and reorganizes them to form regions of interest.Finally,the optimized YOLOv4 network is used to detect birds in the region of interest.Experiment results show that the proposed method can accurately detect birds in substations and has strong practicability.
作者
苏慧民
刘泽
朱文明
章倩
周清楷
王庆华
SU Hui-min;LIU Ze;ZHU Wen-ming;ZHANG Qian;ZHOU Qing-kai;WANG Qing-hua(State Grid Jiangsu Electric Power Co.,Ltd.,Changzhou Power Supply Branch,Changzhou 213000,Jiangsu Province,China;College of Internet of Things Engineering,Hohai University,Changzhou 213022,Jiangsu Province,China;Changzhou Zhongneng Power Technology Co.,Ltd.,Changzhou 213000,Jiangsu Province,China)
出处
《信息技术》
2021年第12期43-48,54,共7页
Information Technology
基金
国网常州供电公司科技项目(SGTYHT/19-JS-218)
江苏省重点研发计划(BE2020092)。
