针对在户外条件下进行图像处理的特点，提出一种适合智能变电站巡检机器人的对变电站设备悬挂异物识别的算法。该算法经过国内某500 kV智能变电站巡检机器人实地反复测试，对有异物与无异物的正负样本图像识别率超过95%，对实时巡检具有高精度高鲁棒性，完全满足变电站现场应用的要求。%A foreign object detection algorithm is proposed,which is adapted to the inspection robot working outdoor in the smart substation. First,the template images are manufactured for electric inspection equipments. SIFT method is utilized to match with the template images for detecting the pending device region from the real⁃time image captured by the robot,and then a normalized method is produced to both device ROI images to reduce the light influence. After the two binary images per⁃formed,an improved Hu moment algorithm is utilized to compare with the difference of the same device. If foreign object is sus⁃pended on the target device,the difference region will be marked in real time by the proposed method. Experimental results show that the proposed method is both effective and accurate.
展开▼
