基于航拍图像的输电线路异物识别

Recognition of Extra Matters on Transmission Lines Based on Aerial Images

为提高无人机电网巡线故障点排查的效率,针对输电线路走廊悬挂异物的潜在故障,提出一种可从大量航拍输电线路图像中筛选出挂有异物的图像的方法.基于形态学算法改进Otsu(最大类间方差)自适应阈值分割算法分割图像背景,提出一种新的基于输电线路特征的滤波方法进一步滤除背景;通过梯度法获取电力线的边缘,选取Hough变换累加器中局部极大值个数与最终检测到的线路数量作为异物特征向量来识别异物.最后,开发出批处理系统识别验证.结果表明,该算法能将挂有异物的电力线图像准确识别,为输电线路可靠性提供保障.

An extra matters recognition method was proposed to sift images with extra matters from amounts of power system images, aiming at improving the detection efficiency on the fault point in transmission line patrol for the potential extra matters danger in the transmission corridor. First, the Otsu self-adaptation threshold segmentation algorithm was improved to segment the aerial images based on the mathematical morphology algorithm. Then a new filter algorithm to extract target image based on the geometrical characteristic of straight line was proposed to filter the rest of background. In addition, the transmission line edges were calculated via gradient. Then the local maxima in Hough transfer accumulator and the final line number were chosen as feature vectors to recognize extra matters. Finally, an interface was developed to verify the algorithm. The result shows that with this method, all the images with extra matters can be sifted, which guarantees the power system.