结合颜色与空间信息的WEB图像检索新方法

A New WEB Image Retrieval Method Combined with Color and Spatial Information

提出一种图像检索新方法。利用HSV颜色模型非等量化实现图像主色的提取以减少WEB图像特征向量维数,再用直方图的交集作为相似性度量进行第一次图像检索;得到检索结果排列靠前在全局颜色分布上相似的部分图像后,进一步考虑颜色的空间分布状况,采用颜色对方法进行二次检索以提高检索精度。

In order to improve the retrieval efficiency,this paper uses the Non-uniform quantization of the HSV color model to extract the main colors of the images,reduces the feature vectors of the WEB images,then uses the histogram intersection as a similarity measure for the first image retrieval.After obtaining the top parts of search results that are similar on the global color distribution,considering the spatial distribution of color,this paper uses the method of colors to improve the retrieval accuracy.Proved by the experiments the method used in this paper can effectively improve the performance of web image retrieval from the efficiency and accuracy.