一种航拍绝缘子串图像自动定位方法

Aerial insulator string image automatic location method

绝缘子串自动定位是实现其运行状态监测与故障诊断的重要前提。针对航拍图像背景复杂的特点,提出一种基于阈值分割和二值形状特征描述的绝缘子串自动定位方法。先利用基于非下采样Contourlet变换(NSCT)的灰熵模型和细菌觅食-粒子群优化(BF-PSO)算法求取阈值,将原始航拍图像分割为二值图像,再利用二值图像中绝缘子串的形状特征,对其特征进行数字化描述,并以此为依据去除非绝缘子目标,实现绝缘子的自动定位。实验结果表明,提出的分割方法可以把目标和背景区分开来,边缘完整、连续,目标内部空洞少,抗噪性好;提出的定位方法可以完成复杂背景下绝缘子串的定位,定位精度高,速度快。该方法无需人工参与,可以为绝缘子的状态监测与故障诊断奠定基础。

Insulator string automatic location is an important premise of its operation condition monitoring and fault detection.Aiming at the characteristics of aerial image complex backgrounds,this paper proposes an insulator string automatic location method based on threshold segmentation and binary shape feature description.The gray entropy model based on non subsampled contourlet transform （NSCT） and bacterial foraging-particle swarm optimization （BF-PSO） algorithm are adopted to obtain the threshold.Then,using the threshold the original aerial image is segmented to a binary image.There are several objects around the insulator string in the binary image.The shape feature of the insulator string is extracted from the binary image,and then the features are described digitally.Based on the binary feature description the phony insulator target is removed and the automatic location of the insulator string is achieved.Experiment results show that the proposed segmentation method can separate the target and background region.The segmented target has complete and continuous edges,has good noise resistance capacity,and in the target there are less holes.The proposed location method can achieve insulator string location under complicated backgrounds with good performance,such as high location accuracy and fast location speed.The proposed method is fullautomatic,and it lays a foundation for insulator condition monitoring and fault detection.