针对电力设备红外图像分割效果受非均匀背景和噪声干扰等因素影响的问题, 提出了一种基于粒子群优化方法的 Niblack设备红外图像分割算法。采用类间方差作为粒子群算法的适应度函数, 自动搜寻 Niblack法中图像不重叠矩形邻域的最优分割阈值, 并将其用于图像的二值化分割, 从红外图像中提取出设备的目标区域。实验结果表明: 该分割算法与传统的 Otsu等算法相比效率更高, 且误分率(ME)减少了 14%～78%。鲁棒性分析表明, 本算法对含较大噪声密度的红外图像分割性能优于其他传统算法, 从而有效提高了电力设备红外图像分割精度与效率。

The infrared (IR) image segmentation effect of power equipment is affected by non-uniform backgrounds, noise interference, among others. To solve this problem, a method of Niblack segmentation algorithm based on particle swarm optimization for infrared images of power equipment is presented. An inter-class variance is adopted as the fitness function of the particle swarm optimization algorithm to automatically search the optimal segmentation threshold value from the non-overlapping rectangular sub-block in Niblack’s method. The device is extracted from the IR image target area. The experiment results indicate that the misclassification rate (ME) of the proposed method is reduced by 14%–78% compared with traditional methods such as Otsu’s. The robustness analysis based on ME demonstrates that the proposed method is superior to other traditional methods in IR images with high noise density. Therefore, the detection accuracy and efficiency of IR image segmentation are improved.