结合Gabor滤波器和ICA技术的纹理分类方法

A Method for Texture Classification by Integrating Gabor Filters and ICA

提取有效的特征用于纹理描述和分类一直是纹理分析的难点.本文提出一种结合Gabor滤波器和ICA技术的纹理特征提取方法,即纹理图像首先经过Gabor滤波器组滤波,然后由滤波图像直接构建高维特征矢量;再将这些高维特征矢量通过主成分分析PCA进行降维,最后采用ICA技术分析和提取降维后的特征矢量中的独立成分用于纹理分类.通过与经典Gabor滤波器和ICA方法的对比实验,验证和评价了本文方法的性能.

Extracting effective features for texture description and classification is always a difficult problem in texture analysis. This paper proposes a method for texture feature extraction by integrating Gabor filters and independent component analysis （ICA）. That is, the texture image is firstly filtered by a given bank of Gabor filters, and then higher-dimensional feature vectors are constructed from the filtered images. Next, the dimensionality of these vectors is reduced by means of principal component analysis （PCA） .Finally, the independent components in the resulting vectors with dimensionality reduced am analyzed and extracted by using ICA for texture classification. Comparative experiments among this approach, the classic Gabor filters and ICA, am performed. The results demonstrate and evaluate its performance.