【摘要】 提出了一种简单的尺度和旋转不变性纹理分类方法 .首先使用对数 -极坐标变换对要分类的各纹理图像进行变换 ,生成对数 -极坐标图像 ;然后以对数 -极坐标图像的正则化以后的行投影数据向量构成纹理图像的特征向量 ;最后运用支持向量机实现尺度和旋转不变性纹理图像的分类 .实验结果表明 ,该方法在尺度和旋转不变性纹理图像的分类中 ,准确率可达 90 %以上更多还原

【Abstract】 A simple and effective texture classification of invariant scale and rotation was proposed. In this classification, firstly, the texture images to be classified are translated into Log-polar images via Log-polar transformation. Then, the feature vectors of texture images are constituted with the normalized row project datum vector of the translated images. Finally, the support vector machines are used to the texture classification of invariant scale and rotation. Experimental results show that the proposed method is of an accuracy of more than 90% when it is used to the texture classification of invariant scale and rotation.更多还原