摘要
提出了一种新颖而又简单直观的图像索引机制。这种方法通过统计图像数据库的低层像素特征,用中值切割法构建索引树,把自然景物图像数据库分成易于描述的不同视觉主题,并籍此来进行高效的图像检索。索引树的构建有别于传统的完全基于数值本身的聚类方法,因而聚类结果视觉含义更明显。本文以颜色主题为例实现了此方法,并通过实验论证了方法的高效和快速。图像索引码以二进制形式给出,因而所需存储空间也极小。
A novel image indexing method is used for fast image querying and browsing. Based on statistics of low level features the approach develops a median cut method to construct an indexing tree. Based on the division of the image database the construction of the indexing tree is divided into intuitive visual themes to improve image indexing. Numerical clustering is difficult to interpret the results visually. This paper presents an implementation of the method based on the color theme division. Experimental results demonstrate that the method is efficient and fast. The index of an image is in the form of a binary key, thus requireing very little sto-rage space.
出处
《数据采集与处理》
CSCD
北大核心
2005年第2期198-202,共5页
Journal of Data Acquisition and Processing
