基于多尺度及多方向分析的纹理图像检索算法Texture Image Retrieval Algorithm Based on Analysis of Multi-scale and Multi-orientation
汪华章
摘要(Abstract):
提出一种基于多尺度及多方向分析的纹理图像检索算法.针对纹理图像的特点,从不同尺度和方向比较了实数离散小波变换和双树复小波变换的性能.充分利用双树复小波变换的旋转不变性、良好的方向选择性以及有限的冗余等优点,将其有效地应用于纹理特征提取过程中.通过提取各子带上的能量和标准差作为特征矢量,采用Camberra距离作为相似度量进行检索,减小了计算量,取得了良好的检索效果.实验结果表明,双树复小波提取纹理特征所获得的检索性能优于实数离散小波检索算法,也优于经典的灰度共生矩阵算法,且算法具有良好的旋转不变性.
关键词(KeyWords): 图像检索;复小波变换;旋转不变性;相似度量
基金项目(Foundation): 四川省科技攻关项目,编号05GG021-026-03
作者(Author): 汪华章
参考文献(References):
- [1]Tuceryan M,Jain A K.Texture Analysis:Handbook Pattern Recognition and Computer Vision[M].Singapor:WorldScientific Publishing Company,1993:235-276.
- [2]Haralick R M,Shanmugam K,Dinstein I.Texture features for image classification[J].IEEE Transactions on Systems,Man and Cybernetics,1973,3(6):610-621.
- [3]刘忠伟,章毓锦.综合利用颜色和纹理特征的图像检索[J].通信学报,1999,20(5):36-40.
- [4]Challa S,Sastry M R,Arun K P,et al.A modified Gabor function for content based image retrieval[J].Pattern Recog-nition Letters,2007,28:293-300.
- [5]Han J,Ma K K.Rotation-invariant and scale-invariant Gabor features for texture image retrieval[J].Image and VisionComputing,2007,25(9):1474-1481.
- [6]Pun C M,Lee M C.Log-polar wavelet energy signatures for rotation and scale invariant texture classification[J].IEEETransactions on PAMI,2003,25(5):590-603.
- [7]Kingsbury N.Complex wavelets for shift invariant analysis and filtering of signals[J].Applied and Computational Har-monic Analysis,2001,10(3):234-253.
- [8]Selesnick I W,Baraniuk R G,Kingsbury N G.The dual-tree complex transform[J].IEEE Signal Processing Magazine,2005,22(6):123-151.