一种基于颜色特征的图象检索方法  

An Image Retrieval Method Based on Color Perceived Feature

作　　者：李国辉[1] 柳伟[1] 曹莉华[1] 

机构地区：[1]国防科技大学

出　　处：《中国图象图形学报（A辑）》1999年第3期248-251,共4页Journal of Image and Graphics

基　　金：湖南省自然科学基金

摘　　要：提出了一种改进的基于颜色感知特征的图象检索方法。这种方法采用符合人类视觉特性的ＨＳＶ颜色空间来表示图象的颜色特性，利用人对颜色的感知来对颜色分量进行非等间隔的量化并形成特征矢量，用直方图交叉算法进行图象的匹配。用户可以通过示例方式、指定颜色或指定图象区域来表示查询的视觉要求。An improved image retrieval method based on color feature is provided. This method uses HSV color space that is more compatible with hum’s visual sense to represent image’s color feature. It takes advantage of human’s perception to quantize color components as non equal interval, which formulates one dimensional feature vector. After that, we use histogram intersection algorithm to search similar images from image repository. Query by example, query by specified color, or query by region within an image is used to represent visual query posed by user. Finally, test results for relevant image retrieval compared with equal quantized arithmetic are given and discussed.

关 键 词：基于内容检索 多媒体数据库 图象检索 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]