您的位置:首页  > 论文页面

输电线路图像绝缘子串实时定位方法

发表时间:2018-11-30  浏览量:70  下载量:16
全部作者: 许周乐,尚赵伟,丁溢洋
作者单位: 重庆大学计算机学院
摘 要: 针对输电线路图像中绝缘子串定位的难点,利用散射变换具有弹性形变稳定性以及卷积神经网络(convolutional neural network,CNN)学习能力强的特点,提出一种速度快、精度高的新方法。首先对图像做散射变换以提取不同尺度、不同方向的系数来抑制光照等的干扰,其次计算低频系数的格拉姆矩阵以增强其边缘纹理特征,再根据训练集数据计算网络特征图尺寸,然后将散射系数通过改进的YOLOv2网络进行初步定位,最后利用集成学习得到精确位置。实验结果表明,该方法在保证实时计算的前提下,与原YOLOv2网络相比,召回率可提升8.46%.
关 键 词: 人工智能;绝缘子;定位检测;散射变换;卷积神经网络;集成学习
Title: Real-time locate method of insulator string in transmission line images
Author: XU Zhoule, SHANG Zhaowei, DING Yiyang
Organization: College of Computer Science, Chongqing University
Abstract: Considering the difficulty of locating the insulator string on transmission line image, in this paper, a fast and accurate novel method based on the elastic deformation stability of scattering transformation and good learning capacity of convolutional neural network (CNN) is proposed. Firstly, scattering transformation is applied on the image to extract coefficients of different scales and directions to suppress the interference of light. To enhance the edge texture features, the Gram matrix low frequency coefficients is calculated. Then, the size of network feature map is calculated according to the training set. With scattering coefficients, the improved YOLOv2 network is trained to generate the initial location of the insulator string. Finally, the accurate location can be obtained by ensemble learning. The experimental results demonstrate that the proposed method can not only improve the recall rate by 8.6% compared with the original YOLOv2 network, but also guarantee the real-time response speed.
Key words: artificial intelligence; insulator; location detection; scattering transformation; convolutional neural network (CNN); ensemble learning
发表期数: 2018年11月第22期
引用格式: 许周乐,尚赵伟,丁溢洋. 输电线路图像绝缘子串实时定位方法[J]. 中国科技论文在线精品论文,2018,11(22):2202-2212.
 
0 评论数 0
暂无评论
友情链接