基于灰度共生矩阵和单尺度MRF的纹理图像分割

为改善纹理图像分割效果，提出一种基于灰度共生矩阵和单尺度MRF的纹理图像分割方法．这种方法考虑到纹理信息在空间内的结构特征以及一个像素与周围像素作用的特性关系，采用灰度共生矩阵的几个二次统计量作为纹理特征向量，利用K-means聚类获得起始分割，然后联合建立MRF的特征场与标号场模型．实验表明，此方法提高了分割准确度与一致性．

Texture Image Segmentation Based on Gray-Level Co-Occurrence Matrix and Single MRF

To improve the texture image segmentation effect, a method of combining the gray-level co-occur- rence matrix and MRF is proposed. By considering the two aspects of spatial distribution characteristics and the relationship between a pixel and the pixels around of the label field from MRF, describing the texture feature based on the gray level co-occurrence matrix of the two statistics and using K-means to obtain initial segmenta- tion,the feature field and the label field of MRF can be established. The experiment results demonstrate that the proposed method can effectively improve the accuracy rate and the consistencv of the segmentation.