基于排列组合熵和灰度特征的纹理分割  

Texture segmentation method based on permutation entropy and gray feature

作　　者：钱诚[1] 范影乐[1] 庞全[1] 

机构地区：[1]杭州电子科技大学模式识别与图像处理实验室,浙江杭州310037

出　　处：《计算机应用》2006年第3期586-588,共3页journal of Computer Applications

基　　金：国家自然科学基金资助项目(60374047);浙江省自然科学基金重点资助项目(ZD0205)

摘　　要：提出了一种基于排列组合熵和灰度特征的纹理分割方法。该方法将不同方向上的排列组合熵与灰度均值、灰度方差结合起来构成一个多维特征向量,利用模糊C均值聚类算法进行聚类实现纹理图像的分割。实验结果表明该方法对纹理分布均匀的图像有着良好的分割效果。在保持较高纹理分割精度的前提下,该方法能减小计算复杂度,并且具有较强的鲁棒性和抗噪声能力。A method for texture segmentation was presented. According to a feature vector which was made up of the permutation entropies calculated in many different directions, gray average and gray deviation, the method made the texture segmentation by fuzzy c-means algorithm. The experiment results show that the method has a good performance on segmenting images with textures distributed uniformly. On the premise of keeping high precision in texture segmentation, the method reduces the computational complexity, and has the ablility of robustness and noise resistance.

关 键 词：纹理分割 排列组合熵 灰度特征 模糊C均值聚类 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]