输电线路覆冰厚度的小波分析图像识别

Wavelet Image Recognition of Ice Thickness on Transmission Lines

针对现有输电线路覆冰厚度检测方法的不足,利用南方电网输电线路灾害(覆冰)预警系统中的覆冰图像,选择3阶B样条函数作为平滑函数构造小波基对覆冰图像进行多尺度边缘检测,再采用Hough变换对图像检测边缘进行连接,通过对比导线覆冰前后的厚度差,实现覆冰厚度自动计算。结果表明:多尺度小波边缘检测信息融合与Hough变换直线检测算法,对离散误差、局部断裂等缺陷不敏感,能剔除覆冰图像中冗余和虚假边缘,从而在实际噪声图像中提取所有覆冰导线边缘,图像分辨率提高有助于减少计算误差。最后,经与南方电网输电线路灾害(覆冰)预警系统监测的数据进行对比,所得误差<2.5 mm,表明该方法可用于输电线路覆冰厚度在线监测。

In allusion to the deficiency of existing methods in ice thickness detection for transmission lines, we proposed a method for realizing automatic thickness detection. Adopting the icing images which were taken from the transmission lines disaster （icing） warning system of China Southern Power Grid, and choosing the 3-order B-spline function as the smooth function, we constructed wavelet bases for multi-scale edge detection of the icing images. Then, we adopted the Hough transform to connect the detected edges of the images, and compared the thickness of the lines before and after ic- ing to realize the automatic calculation for icing thickness. The results show that the information fusion of multi-scale wavelet edge detection and Hough transform line detection algorithm are insensitive to the drawback of discretization er- ror and local fracture, etc., so the redundancy and false edge in icing images can be removed and all the edges of the icing lines can be extracted from actual noise images. Moreover, improved image resolution also decreases calculation errors. Finally, the calculation results are compared with the monitoring data from the warning system and the error is less than 2.5 mm, showing that the proposed method is applicable for on-line monitorinz of ice thickness.