基于模糊集和k-means算法的变压器红外图像分割

摘要

针对k-means算法的k值选定和复杂背景下红外图像误分割问题,提出了一种结合模糊集理论和k-means算法的改进方法.该方法根据灰度级直方图估计k值,在获得k值的基础上,利用直方图均衡化和模糊集理论进行图像增强,然后通过k-means算法结合数学形态学的开运算,再进行图像分割.实验结果表明,该方法获得了更为准确的聚类结果,同时实验对比发现该方法相较其它方法分割效果更好,又兼顾了快速性和变压器温度细节表现能力.%In this paper,an improved method based on fuzzy set theory and k-means algorithm is proposed to solve the problem of the k value and the improper segmentation of infrared image under the complex background.The k value can be estimated according to the gray level of histogram.On the basis of the k value,image enhancement is carried out by the histogram equalization and fuzzy set theory.Then the image is segmented through the combination of k-means algorithm and opening arithmetic of mathematical morphology.The results show that the proposed method not only is more accurate than other methods,but also has better segmentation results and takes into account rapidity and temperature detail performance of transformers.