【摘要】 提出了一种基于显著兴趣点的图像检索新方法,此方法主要有显著兴趣点检测、基于显著兴趣点的特征描述和相似性度量三个步骤.先使用一个自适应滤波器对图像进行滤波,然后提取显著兴趣点;以显著兴趣点为线索,设计了一种基于显著兴趣点的环形颜色直方图,既利用了显著兴趣点的局部特征,又考虑了显著兴趣点的空间分布结构;用图像间的环形颜色直方图距离来度量图像间的相似性.该检索算法不但保证了对图像旋转、平移鲁棒性,而且克服了传统直方图没有空间位置的缺陷.实验结果表明,该方法对图像检索非常有效.更多还原

【Abstract】 A novel algorithm for image retrieval based on prominent interest points is presented. The algorithm consists of three steps: prominent interest points detection, feature description based on prominent interest points and similarity measurement, image is filtered by a self-adaptive filter, then prominent interest points are extracted; prominent interest points are regarded as clues, and annular color histogram is designed, which takes not only the local color feature into consideration but also the space distribution information of prominent interest points; the similarities are measured by the distance of annular color histograms.With robustness to rotation and translation, the algorithm avoids shortcoming of losing the location information in color histogram.Experimental results show that this algorithm is efficient for image retrieval.更多还原