为进一步进行纹理特征分析，从纹理的方向性人手，给出了纹理方向的数学定义式，合理选择差异函数，

构造了具有物理意义的纹理方向描述特征向量．数据处理方面，运用模糊贴近度的概念，结合改进后的属性均值聚

类算法，对一类具有方向性的纹理图象进行分类与分割实验，取得了较好的结果．试验表明，该方法对纹理的方向

性有很好的描述能力．

关键词 图象分割 纹理方向 纹理分割 神经网络 模糊聚类

-Texture features for further analysis, staff from the texture direction, given the direction of the mathematical definition of texture type, a reasonable choice of the difference function, structure with a physical meaning to describe the direction of the texture feature vector. Data processing, the use of the concept of fuzzy nearness degree, combined with improved properties means clustering algorithm for a class of directional texture image classification and segmentation experiments, achieved good results. Tests show that the method of directional texture has a good descr iption of the capacity. Keywords Image segmentation, texture segmentation texture direction fuzzy clustering neural network