基于RGB边缘和颜色特征的图像检索  

Image retrieval based on RGB edge and color

作　　者：台德艳[1] 

机构地区：[1]安徽师范大学物理与电子信息学院,安徽芜湖241000

出　　处：《微计算机信息》2010年第26期208-209,202,共3页Control & Automation

摘　　要：文章提出了一种基于RGB向量空间图像边缘检测和分块颜色直方图熵值的图像检索方法。该方法利用RGB边缘检测得到形状特征,并根据图像主要对象所占图像比例值来调整颜色分块的数目和权值,使用形状和分块直方图熵值的特征矢量加权距离计算图像之间的相似度。实验表明该检索方法效果优于单独特征的图像检索。A new image retrieval algorithm based on image edge detection in RGB color space and color histogram entropy in each sub-block is proposed. Shape features are extracted by image edge detection. Sub -block number and weight value depends on the percentage of the main target in image. The similarity between color images is computed by using a combined feature index based on shape and color histogram entropy. Experimental results show this system is surpassed the single image retrieval.

关 键 词：图像检索 边缘检测 直方图熵值 特征矢量提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]