两种基于空间域聚类分析的彩色图像分割方法比较

A Comparison Between Two Methods of Color Image Segmentation Based on Cluster Analysis in Space Domain

综合研究了基于空间域聚类分析的图像分割方法 ,针对彩色显微颗粒图像特征 ,选择了两种阈值分割法———最大类间方差分割法和模糊聚类分割法 ,并对其进行了推广以应用于彩色图像分割。在详细分析各种彩色特征的前提下 ,通过选择合适的正交彩色特征量I1、I2和I3,成功地进行了两类彩色显微颗粒图像分割和目标提取。算到表明 :两种方法的提取精度相当 ,而最大类间方差法的计算速度明显要快 。

This paper comprehensively studies all the image segmentation methods by thresholding In view of the color microscope particle image characters,the Ostu method with maximum variance between two classes and the fuzzy cluster method are selected and extended to segment color image and abstract color particle objects.After a careful analysis of the various color characters,two kinds of color microscope particle image are segmented successfully through selecting the correct color character I 1,I 2 or I 3.The example shows both methods are equal in abstracting precision,but the Ostu method is much faster and more effcient than the fuzzy cluster method.