基于改进Canny算法的绝缘子裂缝识别

Insulator crack recognition based on improved Canny algorithm

针对传统绝缘子裂缝识别算法存在的不足,提出改进的Canny算法进行绝缘子裂缝识别。采用自适应中值滤波代替高斯滤波对裂缝图像进行平滑处理,在计算梯度幅值上增加45°和135°方向模板,降低错检率,减少边缘丢失现象,最后通过采用最大类间方差法自适应选取图像灰度的高、低阈值。对传统Canny算法和改进后的Canny算法进行了对比实验,结果表明,改进后的Canny算法自适应性强,产生的毛刺较少,裂缝故障识别和定位准确率较高,验证了方法的有效性。

Aiming at the shortcomings of traditional insulator crack identification algorithms,an improved Canny algorithm is proposed for insulator crack identification.Adaptive median filter is used instead of Gaussian filter to smooth the crack image,and 45°and 135°direction templates are added to calculate the gradient amplitude,so as to reduce the error detection rate and the edge loss,and finally the maximum between-class variance method is adopted to select adaptively the high and low thresholds of image gray.A comparative experiment is carried out between the traditional Canny algorithm and the improved Canny algorithm.The results show that the improved Canny algorithm has strong adaptability,produces fewer burrs,and has a higher accuracy of crack fault identification and location,which verifies the effectiveness of the method.