考虑到纹理的方向及相位特性，提出一种基于双树轮廓波变换的纹理图像分割新方法.双树轮廓波由塔型双树方向滤波器组构建，它不仅保持了轮廓波灵活的方向选择性，克服了传统轮廓波不具有移不变性的缺点，而且提供了复子带系数.本文以各方向子带的能量作为方向特征，复系数的平均相角作为相位特征，生成特征向量，利用模糊C均值聚类算法进行纹理图像分割.实验结果表明，相位特征是方向特征的有效补充，与传统方法相比，该方法在分割错误率、边缘准确性以及区域一致性等方面均有明显的改善.

Inspired by the directional and phase characteristics of textures，a novel texture image segmentation approach based on dual tree contourlet transform is proposed.Dual tree contourlet is constructed by the pyramidal dual-tree directional filter bank,it not only maintains the flexibility direction selectivity of the contourlet transform,overcomes the limitation of traditional contourlet which namely lack of shift invariance but also provides complex subband coefficients．The proposed scheme using the energy of each directional subbands as the directional features,and defining the mean phase angle of complex coefficents as the phase features for segmentaion.Finally,the fuzzy C-means clustering method is used to complete the texture image segmentation．Experiment results demonstrated,the phase features are complementary to the directional features; compared with a typical traditional method，the present approach shows visible improvements both in diminishing segmentation error，and in increasing boundary precision and region harmony.