基于排列组合熵和灰度特征的纹理分割.pdf

第 26卷第 3期 计算机应用 V ol. 26 No. 3 2006年 3月 ComputerApplications Mar. 2006 : 1001- 9081( 2006) 03- 0586- 03 钱 诚, 范影乐, 庞 全 (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) ( qcshougao@ tom. com ) :提出 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 : 纹理分割; 排列组合熵; 灰度特征;模糊 C 均值聚类 : P391. 41 : A T exture segm en tation m ethod based on perm u tation en tropy and gray feature QIAN Cheng, FAN Y ingle, PANG Quan (Laboratory of Pattern Recognition and Imagep rocessing, H angzhou Dianzi University, H angzhou Zhejiang 310037, hina) A bstract: A method for texture segm entation w as presented. Accord ing to a feature vector w hich w as m ade up of the perm utation entrop ies calculated in m any d ifferent d irections, gray average and gray deviation, the methodm ade the texture segm entation by fuzzy cm eans algorithm. he expermi ent results show that them ethod has a good perform ance on segm enting mi agesw ith textures d istributed uniform ly. On the prem ise of keep ing high precision in textu re segm entation, the method reduces the computational comp lexity, and has the ablility of robustness and noise resistance. K ey w ords: texture segm entation; perm utation entropy; gray feature; fuzzy cm eans cluster , 0 引言 , ,, [ 1] [ 2] Gabor , : n ,