引用本文: | 朱惠玲,牛哲文,黄克灿,唐文虎.基于单阶段目标检测算法的变电设备红外图像目标识别及定位[J].电力自动化设备,2021,41(8): |
| ZHU Huiling,NIU Zhewen,HUANG Kecan,TANG Wenhu.Identification and location of infrared image for substation equipment based on single-stage object detection algorithm[J].Electric Power Automation Equipment,2021,41(8): |
|
摘要: |
针对红外图像中变电设备的识别和定位问题,提出了一种基于改进YOLOv3算法的变电设备检测方法。在现场采集的变电设备红外图像集的基础上,首先使用基于Retinex的图像增强算法以及阈值分割等图像处理方法对图像集进行预处理;然后基于变电设备红外图像对YOLOv3算法进行参数优化,并通过迁移学习的策略对改进YOLOv3网络进行训练以解决图像集样本数量较少的问题。实验结果表明,在样本数量较少的情况下,所提方法可以达到满意的检测准确率,并能快速地实现变电设备的识别和定位。 |
关键词: 变电设备 目标检测 Retinex图像增强 YOLOv3 迁移学习 |
DOI:10.16081/j.epae.202104015 |
分类号:TM507 |
基金项目:国家自然科学基金资助项目(51977082);中央高校基本科研业务费专项资金资助项目(x2dl-D2181850);广东电网有限责任公司科技项目(031800KK52180081) |
|
Identification and location of infrared image for substation equipment based on single-stage object detection algorithm |
ZHU Huiling, NIU Zhewen, HUANG Kecan, TANG Wenhu
|
School of Electric Power Engineering, South China University of Technology, Guangzhou 510006, China
|
Abstract: |
In order to identify and locate the substation equipment in infrared image, a detection method based on improved YOLOv3 algorithm is proposed. Using the infrared image set of substation equipment collected in the field, the image enhancement algorithm based on Retinex and image processing methods such as threshold segmentation are used to preprocess the image set. Then the parameters of YOLOv3 algorithm are optimized based on infrared images of substation equipment, and the improved YOLOv3 network is trained through the transfer learning strategy to solve the problem of insufficient number of image set samples. The experimental results show that the proposed method can achieve a satisfactory detection accuracy in the case of a small number of samples, and can quickly identify and locate substation equipment in infrared images. |
Key words: substation equipment object detection Retinex image enhancement YOLOv3 transfer learning |