Abstract: This paper presents a segmentation algorithm for infrared image of polluted and wetted insulators.First，it adopts a window gray stretching method to intensify the image，and then，binarization process is done to the image.After that，it uses morphology algorithm to combine the correlated zones and split the uncorrelated zones of the image.Finally，it employs edge configuration scanning method based on gradient to remove conducting wires in the image.Simulation results show that the method can divide the polluted and wetted insulator from the background efficiently with sharp edge.It lays a foundation for further feature extraction and contamination degree fixed.

摘要： 提出了一种湿污绝缘子红外图像的综合分割算法，首先用窗口灰度拉伸算法对图像进行增强，接着对图像进行二值化；然后用形态学方法将图像中相关的区域连接起来，将不相关的区域分离开；最后用基于梯度的边缘轮廓扫描法对获得的图像进行处理，将绝缘子图像中的导线去除。仿真结果表明，该分割方法可以很好地将湿污绝缘子从背景上分割出来，边缘清晰，分割质量良好，为后续特征提取和污秽等级的确定奠定了基础。