摘要
在图像各种形变条件作用下,传统的匹配算法所提取的图像特征匹配度不高,导致对GIS设备内部发热的判断并不准确,为此研究SIFT特征点匹配算法在GIS设备内部发热判断中的应用。此过程中,根据GIS设备内部结构特点,利用二维离散高斯卷积构建GIS设备内部尺度空间。通过该空间的差分检测函数,对尺度空间内的所有极值点进行过滤,以求导的方式得出空间内的关键极值点,从而确定可提取特征点的所在位置。计算梯度模值和梯度方向,生成特征描述子,实现对特征点的详细、分区描述,计算相邻特征点之间的比率,以特征点之间的欧式距离匹配特征,实现对GIS设备内部发热情况的判断。实验结果表明:与传统的匹配算法相比,SIFT特征点匹配算法检测到的图像发生旋转时,所提取的特征点不会减少、其特征匹配度更高。由此可见,该算法更适用于判别GIS设备内部发热的情况。
Under the influence of various deformation conditions,the image features extracted by the traditional matching algorithm are not highly matched,resulting in inaccurate judgment of internal heating in GIS equipment.Therefore,the application of SIFT feature point matching algorithm to internal heating in GIS equipment is studied.In this process,according to the internal structure characteristics of GIS equipment,two-dimensional discrete gaussian convolution is used to construct the internal scale space of GIS equipment.Through the differential detection function of the space,all extremum points in the scale space are filtered,and the key extremum points in the space are obtained by means of derivation,so as to determine the location of feature points that can be extracted.Calculating the gradient mode value and gradient direction,to generate feature descriptors,the detailed and divisional description of feature points are realized.Calculating the ratio between adjacent feature points,and use the Euclidean distance between feature points is used to match features,so as to realize the judgment of internal heating of GIS equipment.The experimental results show that compared with the traditional matching algorithm,SIFT feature point matching algorithm can not reduce the extracted feature points and has higher feature matching degree when the detected image rotates.It can be seen that this algorithm is more suitable for identifying internal heating of GIS equipment.
作者
张平
李东民
苏晓宇
李冠群
朱光南
ZHANG Ping;LI Dongmin;SU Xiaoyu;LI Guanqun;ZHU Guangnan(National Power Network Jibei Electric Power Co.,Ltd.Zhangjiakou Power supply Co.,Ltd.,Zhangjiakou Hebei 075000,China;Zhejiang Dali Science and Technology Co.,Ltd.,Hangzhou 310053,China)
出处
《自动化与仪器仪表》
2020年第7期190-193,197,共5页
Automation & Instrumentation
基金
国家电网公司科技资助项目(No.2018YD56-21)。
