提出一种利用Brushlet变换分解图像,提取纹理特征进行图像检索的新方法.该方法利用Brushlet的特性,成功地检测到了纹理图像的方向信息,弥补了小波变换的不足,从而形成能够精确表征图像的特征矢量.匹配算法则使用比值相似度.实验结果表明,该方法计算简单、有效、检索效果较好.%A new method for image retrieval is presented which decomposes image and extracts image' s texture features using Brushlet transformation. Making use of the advantage of Brushlet, the method successfully detected the directional information of texture image, which makes up the deficiency of wavelet transformation. The method forms feature vectors that can represent image accurately. Besides,the ratio similarity is adopted in match algorithm. Experimental results have shown that the method is simple in computation and is effective, and achieves satisfied retrieval results.
展开▼