基于颜色特征的图像摘要算法  

Image Hashing Method Based on Color Feature

作　　者：赵琰[1] 周俊杰[1] 

机构地区：[1]上海电力学院计算机与信息工程学院,上海200090

出　　处：《上海电力学院学报》2010年第5期489-492,共4页Journal of Shanghai University of Electric Power

摘　　要：提出了一种基于颜色特征的图像摘要算法.首先将原始图像的RGB颜色模型转换到HSI模型中进行量化并转换为一维矢量,然后将图像分为64块,取得每个块内颜色的特征均值,最后进行量化产生固定长度的图像哈希序列.实验结果表明,相似内容的图像哈希距离较小,而不同内容的图像哈希距离较大.该算法还具有一定的篡改检测能力.An image hashing method based on color features is proposed.Firstly,the RGB color model is modified to HIS model.Then the color features are quantified to one dimension vector.The image is split to 64 blocks and the mean color feature is obtained.Finally,the image hashing value is obtained by quantifiing the mean color feature.The experimental results show that the distance between similar images is small,while the distance of different image pair is big.At the same time,this method can detect the color tampering.

关 键 词：图像摘要 颜色特征 色度 饱和度 亮度 

分 类 号：TN911.73[电子电信—通信与信息系统]