摘要： 针对基于文本的图像检索方法检索精确度不高的问题,提出一种新的图像检索方法。为更细致地描述颜色差异,对HSV颜色空间的色调和饱和度分量进行合理量化。在亮度分量上提取局部对角色差描述符,表示颜色的空间分布信息。对综合颜色特征和颜色空间分布信息进行检索,解决颜色直方图不能描述颜色空间分布信息的问题。选取Corel 图像库进行检索和对比分析,实验结果表明,该方法具有较高的检索精确度,同时缩短了图像检索时间。

关键词: 图像检索, 局部对角色差, HSV颜色空间, 颜色直方图, Corel图像库

Abstract: For retrieval accuracy of Text-based Image Retrieval(TBIR) method is not high,this paper proposes a new image retrieval method.The hue and saturation components are reasonably quantified in HSV color space to make more detailed description for color difference.Local Diagonal Color Difference(LDCD) descriptor is extracted on the luminance component to reveal the spatial distribution information of color.Composite color feature and color spatial distribution information are retrieved,which solves the problem that color histogram cannot describe the color spatial distribution information.Experimental results show that this method can achieve higher retrieval accuracy and reduce the time required for image retrieval by selecting the Corel image database.

Key words: image retrieval, Local Diagonal Color Difference(LDCD), HSV color space, color histogram, Corel image library

中图分类号: 

• TP391.41