绝缘子图像分割是通过图像处理技术实现绝缘子识别和提取的基础性操作。为了能够准确地分割出绝缘子图像,提出一种基于改进型单位连接脉冲耦合神经网络(UL-PCNN)的绝缘子图像分割算法。根据相邻神经元之间的关系,改进原始UL-PCNN模型中的连接输入和耦合系数;利用改进的UL-PCNN模型对绝缘子图像进行分割,得到多幅输出图像;利用梯度算法计算原始图像和输出图像的边缘,并分别计算输出图像和原始图像边缘的均方误差(MSE),均方误差值最小的输出图像即为分割效果最好的绝缘子图像。实验结果表明,本文算法能够准确地分割出不同环境下的绝缘子图像,并具有较好的抗噪性能。

Insulator image segmentation is the basic operation used to conduct insulator recognition and extraction using image processing. To segment insulator images accurately, an improved unit-linking pulse-coupled neural network (UL-PCNN)-based insulator image segmentation method is proposed in the present study. First, the link input and coupled parameters of the original UL-PCNN model are improved based on the relationship between a neuron and its neighbors. Next, the improved model is used to segment an insulator image to obtain multiple output images. Finally, the gradient algorithm is used to calculate the edges of the original image and output images, and the mean square error (MSE) of the edge of the original image and MSE of each output image are calculated. The output image with the smallest MSE is considered as the optimal result of insulator image segmentation. The experimental results demonstrate that this improved method can accurately segment insulator images in different environments and has good anti-noise performance.