Abstract: This paper proposes a simple and effective adaptive unsupervised method. Fuzzy Histon threshold technique is applied to obtain all possible homogeneous regions of the color image in the Lab space. Finding the peak value, regions initialization, color on the regional similarity combination, it obtains the homogeneous regions labeling by the corresponding cluster centers. An adaptive FCM clustering algorithm is used to improve the compactness between the homogeneous regions in the final color segmentation. This algorithm has been successfully applied to the Berkeley image database, compared with some current segmentation algorithms, such as： Mean-Shift and NCuts make a reasonable division of better, in the vision it effectively extracts the target, has certain robustness.

Key words: color segmentation, Lab even color spaces, fuzzy Histon-histogram threshold, Fuzzy C-Means（FCM） clustering

摘要： 提出了一种简单有效的自适应无监督方法。在CIELab空间中利用模糊Histon阈值技术获得图像中所有可能的均匀区域，即通过寻找峰值，区域初始分割和区域颜色相似性合并，获得由聚类中心标注的均匀区域，提出自适应FCM聚类算法以提高均匀区域之间的紧密度，最终完成色彩分割。该算法已成功应用到伯克利图像库，相比当前一些无监督色彩分割算法，例如：Mean-Shift、NCuts取得了合理更好的划分，视觉上有效提取目标物体，具有一定鲁棒性。

关键词: 色彩分割, Lab均匀色彩空间, 模糊Histon直方图阈值, 模糊C均值（FCM）聚类