改进FCM聚类在彩色荧光图像分割中的应用

Application of Improved FCM in Color Fluorescence Image Segmentation

在石油地质研究中,荧光图像分割是岩石薄片沥青组分分析的关键步骤。根据不同沥青组分颜色不同的特点,提出一种基于改进模糊C均值聚类(FCM)的彩色荧光图像快速分割方法。首先在RGB空间对荧光图像中的颜色利用人眼能识别的最小颜色差异(JND)概念进行量化,然后将量化后的颜色转换至HLS颜色空间,对HLS颜色空间中的奇异点和非奇异点分别进行FCM聚类。同时为了防止FCM聚类陷入局部最优的情况,对初始聚类中心的选择也进行了优化。实验证明,与传统FCM彩色图像分割相比,论文方法对于荧光图像的分割有更好的效果,在分割速度上面也有很大提高。

In petroleum geology studies ,fluorescence image segmentation is a key step for asphalt component analysis of rock slices .A segmentation method based on fuzzy C‐means clustering（FCM ） for fluorescence color image is proposed .In the method colors of image are quantified on RGB color space before converted to HLS color space .Then the colors on HLS space are classified as non‐singular points and singular points ,and are clustered using FCM respectively .Meanwhile ,a opti‐mized method of initializing cluster centers is used to prevent the FCM clustering into local optimum circumstances .Experi‐ments show that ,compared with the traditional FCM color image segmentation ,this method makes better effect on fluores‐cence image segmentation ,and the speed of segmentation has also improved greatly .