面向大规模图像库的高维索引机制研究

Research on High-dimensional Indexing Scheme for Large Image Database

针对图像的72维HSV颜色特征,提出了一种新的降维索引方法.区别于传统的降维机制,该方法在降维的过程中不仅保留了原始数据空间整体的重要信息,也准确抓住了高维个体数据的重要特性.在大规模图像库上的实验表明,基于本文索引机制的搜索算法不仅显著减少了支配检索时间的I/O开销,而且具有较高的查询准确率.

Aiming at the 72-dimensional HSV feature of image color, this paper proposes a new dimension reduction indexing method. Compared with other dimension reduction schemes, this method can not only retain the important information of the whole data space, but also preserve the important merits of high-dimensional individuals. The experiments on a large image database show both a remarkable reduction of I/O overhead of the data accesses which dominates the query time in the searches and a remarkable improvement performance in the searching.