基于Canny算法的10kV电缆主绝缘表面缺陷识别方法

主绝缘表面存在划痕与污渍类施工缺陷容易导致 绝缘故障,采用图像识别技术可以 智能管控施工缺陷。针对10 kV电缆主绝缘表面划痕与污渍缺陷边缘 不易检测的问题,本文提 出一种基于Canny算法的10 kV电缆主绝缘表面缺陷识别方法。本文首 先利用尺寸不变特征变换(scale-invariant feature transform,SIFT)方法完成对 电缆主绝缘图像的拼接,再借助Canny算法计算主绝缘表面缺陷边缘的梯度与幅值、抑制极 大值信息点、去除伪边缘,得到由划痕与污渍产生的缺陷边缘信息。实验结果表明,由本文 算法检测后电缆主绝缘表 面缺陷图像的边缘轮廓具有较高的纯净度和完整性,缺陷边缘更加 清晰,算法平均准确率达到95.42%,验证了算法的有效性。

Defect identification method for main insulation surface of 10kV cable based o n canny algorithm

The construction defects such as scratches and stains on the main insu lation surface can easily lead to insulation faults.The construction defects ca n be intelligently controlled by using image recognition technology.To solve th e problem that it is difficult to detect scratches and stains on the main insula tion surface of 10kV cable,this paper proposed a defect recognition algorithm f or the main insulation surface of 10kV cable based on Canny operator.In this pa per,the scale-invariant feature transform (SIFT) method is firstly used to complete the stitching of the cable main insulation images to obtain a complete original image,using a gaussian filter to remove the image noise.And then the Canny operator is used to calculate the gradient and amplitude of the defect edge on the main insulation surface,suppre ss the maximum information point,remove the false edge,and get the defect edge information generated by scratches and stains.The experimental results clearly show that the edge contour of the defect image of the cable main insulation sur face detected by the Canny algorithm in this paper has higher purity and complet eness,the defect edge is clearer,and the average accuracy of the algorithm rea ches 95.42%,which verifies the effectiveness of the algorithm.