基于颜色-空间二维直方图的图象检索  

Image Retrieval Based on Color-Space2D Histogram

作　　者：刘芳[1] 王涛 周登文 

机构地区：[1]清华大学计算机科学与技术系国家CAD工程中心,北京100084

出　　处：《计算机工程与应用》2002年第12期85-88,共4页Computer Engineering and Applications

摘　　要：基于颜色直方图特征的图象检索方法简单、高效,但丢失了颜色的空间分布信息。文章给出了一种基于颜色-空间二维直方图进行图象检索方法。该方法使用HSV颜色模型将颜色空间压缩为32种具有代表性的颜色,设计了一种具有重叠方式的图象分块方法,然后统计每块区域内的颜色直方图得到该图象的颜色-空间二维直方图,并给出了相应的相似度计算方法。实验表明该方法实现简单,能够更加灵活、准确地描述图象的颜色特征,从而有效提高了检索的精度。The paper gives an image retrieval method based on color-space histogram.First,it uses HSV color model to compress color space to32kinds of representative colors,then designs8overlapped sub-regions to get color histogram in each region,finally the similarity is estimated based on the color-space two dimension histogram.The method is sim-ple and effect which avoid the shortcoming of losing the space distribution information in color histogram in some ex-tent and the trouble in getting the space distribution information by image segmentation.The method is more flexible and accurate to describe the color feature of an image and improves the image retrieval precision.

关 键 词：颜色-空间二维直方图 图象检索 相似度 图象识别 计算机 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]