基于轮廓波的纹理图像特征提取方法

Contourlet-based Feature Extraction for Texture Images

本文利用轮廓波分解的方向性和能量变化特性,提出了一种适用于纹理图像的特征提取方法。该方法首先通过方向可控滤波器(Steerable Filter)获得纹理图像的角度信息,然后进行轮廓波分解提取纹理在不同分解方向上的特征向量,最后将纹理的角度和方向分解的能量信息一起作为纹理的特征。由于轮廓波的方向和分解层次数目可以灵活地调节,因此特征值数量也是调节的,以达到最好的识别分类效果。本文利用简单的距离分类器对所提出的算法进行了验证,以Brodatz纹理库中的旋转纹理图像为标准,纹理分类的结果达到了满意的效果。

According to the directional property and coefficients energy feature in contourlet decomposition,we proposed a new algorithm which is adapted to extract the texture＇s features.The proposed method gets the directional information by steerable filter and then gets texture features by contourlet decomposition.At last the directional feature and contourlet coefficients features are combined as texture＇s feature.Because the number of contourlet decomposition level and directions can be easily modified,the number of features can be adapted to different applications.With the extracted feature vectors,we can easily distinguish different texture in standard Brodatz texture database with high classification accuracy.