基于向量模糊模式识别的纹理图像分割方法

An Approach to Image Texture Segmentation Based on the Theory of Vector Fuzzy Pattern Recognition

提出了一种基于同步自回归（ＳＡＲ） 模型和模糊信息原理进行纹理分割的方法．利用二阶ＳＡＲ模型对图像纹理进行描述，用最小平方误差方法对模型参数进行估计，将估计的参数作为纹理的特征向量．根据模糊信息原理，分析了纹理图像的模糊特性，给出了一种基于向量模糊集合簇模糊模式识别的纹理分割方法．实验结果表明，用所提出的方法进行图像纹理分割取得了较好的分割效果．

Texture segmentation approach based on simultaneous auto regressive (SAR) model and the theory of fuzzy set is presented. Image texture is described with the second order SAR model and the least square error estimation is employed to estimate the model parameters, which will be used as the characteristic vector of the texture. According to the theory of fuzzy set, the characteristic of fuzzy information of texture image is analyzed, and the approach to the mentation of texture image based on the vector fuzzy pattern of recognition theory is given. Experimental results show that better segmentation results are obtained by the method than those by the traditional Euclidean distance method.