综合利用目标区域颜色纹理特征的彩色图像检索  

Color Image Retrieval Using Both Color and Texture Features in Object Regions

作　　者：刘宁钟[1] 杨静宇[1] 

机构地区：[1]南京理工大学计算机系,南京210094

出　　处：《计算机工程与应用》2003年第32期187-189,203,共4页Computer Engineering and Applications

摘　　要：基于内容的图像检索是当前图像数据库领域的一个研究热点。文章提出了一种基于目标区域的多特征彩色图像检索算法。首先获取彩色图像HSI色彩空间的亮度分量灰度图像,再计算得到亮度分量灰度图像的二值边缘图像,并根据边缘图像的连通性提取彩色图像的目标区域。然后抽取目标区域的颜色特征和纹理特征,并分别对两个特征的相似距离进行归一化。最后综合利用这两个特征共同进行检索。实验结果表明,该文的算法具有良好的性能,在检索准确率上获得了较好的效果。Content Based Image Retrieval is an important and fascinating point of the image database.A color image re-trieval algorithm based on multiple features in object regions is proposed in this paper.First,intensity component grayscale image in the HSI color space of the color image is obtained,and then the binary edge image of intensity component grayscale image is calculated.The object regions are extracted from the color image based on the connectivity of the edge image.Then we obtain the color and texture features of object regions and normalize them separately.Final-ly,we combine the color feature and texture feature for image retrieval.The experimental results show that the algorithm can produce good performance in terms of the retrieval effect.

关 键 词：彩色图像检索 目标区域 颜色 纹理 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]