EDLine与改进光流法线路异物智能检测

EDLine and improved optical flow method for intelligent invasion detection on power lines

为提高无人机电力故障巡检效率,提出一种结合边缘绘制线(edge drawing line,EDLine)特征提取算法与改进光流法的输电线路异物智能检测方法。鉴于背景纹理对航拍图像的影响,研究一种基于矩阵的线段融合算法以解决EDLine线段检测的断续问题,根据多特征约束提取输电线路区域。为有效识别异物,利用动量因子改进光流算法迭代收敛速度,根据光流信息检测异物故障。对无人机巡检图像的测试结果表明,该方法在复杂背景或异物不明显时均可有效检测出异物。提出检测算法准确度高、鲁棒性强,保证巡检的时效性。

To improve the efficiency of power failure inspection for unmanned aerial vehicles,an intelligent detection method was proposed for invasion on transmission lines combining EDLine(edge drawing line)line feature extraction and improved optical flow method.In view of the influence of background texture on aerial images,a matrix-based line segment fusion algorithm was studied to solve the intermittent problem of EDLine detection,and the region of transmission lines was extracted according to multi-feature constraints.To identify the foreign objects on the wire,the momentum factor was used to improve the iterative convergence speed of the optical flow algorithm,and the foreign object faults were detected based on the optical flow information.The test results of UAV inspection images show that the method can effectively detect foreign objects in a timely and effective manner under the complex background and when the foreign objects are not obvious.The proposed foreign body detection method has the advantages of high accuracy and strong robustness,and guarantees the timeliness of inspection.