基于颜色矢量角的彩色图像检索算法研究  

Studies of Color Image Retrieval Based on Color Vector Angle

作　　者：赵珊[1] 安志勇[1] 周利华[1] 

机构地区：[1]西安电子科技大学多媒体研究所,陕西西安710071

出　　处：《红外技术》2006年第8期460-465,共6页Infrared Technology

摘　　要：通过分析利用颜色直方图进行图像检索时存在的问题,提出了一种基于图像边缘空间分布特征的图像检索算法。即利用颜色矢量角对亮度不敏感,对色度和饱和度敏感的特性来提取图像的边缘信息。在图像的边缘空间,提取图像边缘点的颜色矢量角直方图,以此来描述图像的内容。该算法充分利用了图像的颜色信息、边缘信息及形状信息,实验表明该算法具有较好的光照不变性、尺度不变性、平移不变性和旋转不变性。Analyzing the content-based image retrieval on histogram, a novel image retrieval method based on the image edge information is proposed. The image edge information is extracted by the color vector angle which is insensitive to illumination but sensitive to hue and saturation. The color vector angle histogram is constructed in the image edge and used to describe the image. This algorithm makes full use of color and edge and shape information. The experimental results indicates that this method has a good robustness against illumination, position and scale translations.

关 键 词：基于内容的图像检索 颜色矢量角 直方图 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]