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.