基于颜色自相关和颜色空间分布熵的图像检索方法  

Image retrieval method based on color correlation graph

作　　者：杨得国 胡少一 冷齐 YANG De-guo;HU Shao-yi;LENG qi(College of Computer Science and Engineering Northwest Normal University,Lanzhou 730070,Gansu,China)

机构地区：[1]西北师范大学计算机科学与工程学院,甘肃兰州730070

出　　处：《西北师范大学学报（自然科学版）》2018年第3期47-50,共4页Journal of Northwest Normal University(Natural Science)

基　　金：国家自然科学基金资助项目(61165002)

摘　　要：颜色特征是图像重要的特征,颜色相关图则是一直以来基于图像内容检索中最常用的特征描述方式,但是目前基于颜色相关图的图像检索算法存在着计算复杂度高,检索精确度低的问题.为了弥补基于颜色相关图的图像检索算法所存在的不足之处,本文从颜色量化上对颜色直方图进行了修改并结合空间分布熵来描述颜色的空间关系.根据8中心色量化规范得出了64中心色量化最终将所有颜色划分为64类,用颜色直方图来描述此64类颜色.空间分布熵则反映了某种颜色的像素在图像空间中的平均分散程度.实验结果表明,将两者结合使用作为颜色特征矢量,不但减少了计算复杂度,而且还提高了检索精度和检索效率.Color-related features are the most important features of image retrieval.Color-related graphs are the most commonly used feature description methods in image content retrieval.However,current image retrieval algorithms based on color correlation graphs have high computational complexity and low search precision.In order to make up for the deficiencies of image retrieval algorithms based on color correlation,this paper modified the color histogram from color quantification and described the spatial relationship of color with the spatial distribution entropy.We obtained the central color quantification based on the 8-center color quantification specification.Finally,all the colors are divided into 64 classes,and the color histogram is used to describe the 64 colors.Spatial distribution of entropy is a reflection of the color of the pixels in the image space,the average degree of dispersion.The experimental results show that the combination of the two as the color feature vector not only reduces the computational complexity,but also improves the retrieval precision and retrieval efficiency.

关 键 词：图像检索 颜色特征 颜色相关图 颜色直方图 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]