基于Hough检测和C-V模型的航拍绝缘子 自动协同分割方法*

绝缘子分割是通过图像处理技术实现其运行状态自动检测及故障诊断的重要前提。针对航拍图像具有背景复杂、分辨率较低、数量多和伪目标多等特点,使用传统分割方法会产生大量的用户交互导致分割效果不佳。本文把协同分割引入到绝缘子航拍图像处理中,提出一种Hough检测修复结合自动初始化轮廓C-V模型的航拍绝缘子图像协同分割方法。本方法利用航拍绝缘子图像帧之间的关系作为先验信息以达到更高的分割精度。首先对航拍图像进行去除文本预处理;然后对预处理过的图像进行Hough检测修复以处理输电线与绝缘子粘连问题并用SLIC进行超像素分割;最后利用广义霍夫变换实现C-V模型初始轮廓的选取并进行基于图像间的C-V模型的绝缘子协同分割。实验结果表明,本文分割方法的准确率明显比其他算法高,能够有效地区分目标和背景并去除杆塔、输电线等伪目标,自动化性能良好,为无人机航拍绝缘子的状态检测及故障诊断奠定基础。

Automatic co-segmentation method for aerial insulator based on Hough detection and C-V model

Insulator segmentation is an important premise of achieving its operation condition automatic detection and fault detection with image processing technique. The aerial images have the characteristics of complex background, low resolution, large number of images, large number of fake targets, and etc. The traditional single segmentation method causes large amount of user interaction and results in bad segmentation quality. This paper introduces co-segmentation into aerial insulator image processing and proposes a co segmentation method for aerial insulator images combining the C V model with automatic initialized contour and Hough detection & inpainting. The method takes the relationship among the image frames of the aerial insulator images as the prior information to achieve the higher segmentation accuracy. First the text information is removed from the aerial image in the preprocessing, and then the Hough detection and inpainting algorithm is conducted on the preprocessed image to solve conglutination problem of the power lines and insulators in the aerial images. Then, SLIC is used to perform the superpixel segmentation of the recovered images. Finally, the GHT is used to carried out the initial contour selection of the C V model and conduct the insulator co segmentation based on the C V model among the images. The experiment results show that the segmentation accuracy for the proposed method is obviously higher than those for the other algorithms. The method has good automation performance, can effectively separate the target from background region, and efficiently eliminate fake targets, such as pole tower and power lines, which provides a basis for the insulator state detection and fault diagnosis with unmanned aerial vehiclel.