基于加权模糊c均值聚类的快速图像自动分割算法  

A Fast Automatic Image Segmentation Algorithm Based on Weighting Fuzzy c-Means Clustering

作　　者：杨润玲[1] 高新波[1] 

机构地区：[1]西安电子科技大学电子工程学院

出　　处：《中国图象图形学报》2007年第12期2105-2112,共8页Journal of Image and Graphics

基　　金：国家自然科学基金项目(60202004);教育部重点项目(104173)

摘　　要：图像分割是指将一幅图像分解为若干互不交迭的区域的集合,是图像处理和计算机视觉的基本问题之一。为了提高图像分割的效率,提出了一种基于2维直方图加权的塔形模糊c均值(FCM)聚类图像快速分割算法。该方法先通过构造合理的2维直方图对噪声进行抑制;然后通过塔形分解来缩减聚类样本集;最后利用加权FCM聚类算法进行分类。仿真结果表明,该方法的效率明显优于标准的FCM算法。此外,为确定分割的最优类别数c,还引入了一种基于该快速算法的聚类有效性评价函数——修正划分模糊度,实现了最佳图像分割类别数c的自动确定。基于人造图像和实际图像的测试实验结果表明该方法是有效的。Image segmentation denotes a process by which a raw image is partitioned into nonoverlapping regions. The image segmentation problem is treated as a key issue in image processing and machine vision. In order to improve the efficiency of image segmentation, a weighting fuzzy c-means clustering（ FCM ） algorithm based on 2D histogram and pyramid decomposing is proposed in this paper. Creating a reasonable 2DH to restrain noise, pyramid decomposing to decrease clustering samples, and using weighting FCM algorithm to classify the image, through which makes the proposed method prior to the standard FCM algorithm. At the same time, for determining the optimal category number c of image segmentation, a cluster validity function modified by partition fuzzy degree is introduced. The method can determine this optimal category number automatically. The experimental results on the synthetic and real images illustrate the effectiveness of the proposed method.

关 键 词：图像分割 加权模糊c均值聚类算法 聚类有效性函数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]