一种融合IFOA和K⁃Means聚类的低照度图像分割方法

Method of low⁃light⁃level image segmentation based on integration of IFOA and K⁃Means clustering

为改进电气化铁路接触网补偿器监测装置在光照不足时对图像目标区域分割精度较低,无法准确识别入侵异物的问题,采用全局自适应色调映射的方法增强低照度图像,联合改进的果蝇算法与K⁃Means聚类算法(IFOA⁃K⁃Means聚类算法)实现目标区域的准确分割。实验结果表明,该方法对退化图像的分割精度更高,能够充分保持图像的边缘信息,运算开销较小,能有效提高图像后续处理的效率。

The monitoring device of electrified railway catenary compensator has low segmentation accuracy of the image target area when illumination is insufficient,so it is impossible to accurately identify the invading foreign objects.In view of this,a global adaptive tone mapping method is adopted to enhance the low⁃light⁃level image,and realize accurate segmentation of target area by combining the improved fruit fly optimization algorithm and the K⁃Means clustering algorithm(IFOA⁃K⁃Means clustering algorithm).The experimental results show that the method based on IFOA⁃K⁃Means clustering algorithm has higher segmentation precision for degraded image,can fully maintain the edge information of the image,has less operation,and effectively improve the efficiency of image subsequent processing.