（华中科技大学 数学与统计学院， 武汉 430074）
摘 要：结合小波变换的多尺度性和Contourlet变换的多方向性，提出了一种新的基于离散平稳小波变换和无下采样方向滤波器组(stationary wavelet transform and nonsubsampled directional filter banks,SWTNSDFB)的纹理分类方法，采用具有平移不变性的离散平稳小波先进行多尺度分解；然后对每层分解得到的高频子带采用非下采样方向滤波器组进行多方向分解，再计算低频子带和各层方向子带的能量作为纹理特征；最后用支持向量机实现纹理分类。实验结果表明，该方法有效地提高了纹理分类的正确率，而且在小样本情况下，依然可以得到较好的结果。
关键词：特征提取; 纹理分类; 基于离散平稳小波变换和无下采样方向滤波器组; SWTNSDFB;支持向量机
Texture classification based on stationary wavelet transform and nonsubsampled directional filter banks
XIE Jianhui, XIE Songfa
(School of Mathematics & Statistics, Huazhong University of Science & Technology, Wuhan 430074, China)
Abstract:This paperproposed a new method for texture classification based on SWTNSDFB,with the combination of multiscale of wavelet transform and multidirection of Contourlet transform.Firstly, performed multiscale decomposition of the source images with stationary wavelet. Also made directional decomposition of the highfrequency subbands in every scale with nonsubsampled directional filter bands. Secondly,took the energy measures of all the subbands including one lowpass band and every directional subband in each level as the texture feature.Finally, used support vector machines to the texture classification. The experiment results show the proposed method can get more results for texture classification.At the same time, in the condition of small scale samples, the method still has very well results.......