基于多尺度Radon变换的图像检索  

Content-Based Image Retrieval Based on the Multi-Scale Radon Transform

作　　者：安志勇[1] 赵珊[1] 王晓华[1] 周利华[1] 

机构地区：[1]西安电子科技大学多媒体研究所,西安710071

出　　处：《光子学报》2007年第6期1176-1180,共5页Acta Photonica Sinica

基　　金：十五国防科技(电子)预研项目(413160501)资助

摘　　要：根据Radon变换的统计特性构造了不变量,提出一种新的基于多尺度Radon变换的图像形状检索方法.对检索图像作小波变换,根据小波模极大原理得到边缘图像,对边缘图像构造Radon变换中心矩,在中心矩的基础上根据Radon的统计原理构造出尺度不变矩.由于矩阵的奇异值具有旋转不变性,因此针对不变矩向量矩阵求奇异值,该奇异值特征向量具有平移、尺度和旋转不变性.将该Radon变换的不变量作为形状特征,并进行高斯归一化,按照欧氏距离计算不同图像间的形状相似度.试验结果表明,该方法对高斯噪音具有较强的鲁棒性,与其它方法相比具有较好的检索效果.The invariant u retrieval algorithm based is achieved by the centra modulus maximum. In th analyzing the statistical sing statistical theory of Radon transform is constructed and a new kind of image on multi-scale Radon transform is presented. In the first place,the shift invariant 1 moment on the Radon transform of edge images,which is gained by the wavelet e second place,the scaling invariant based on the central moment is deduced after characteristics of Radon transform. Since the singular values of matrix are uncorrelated with the positio the rotational invariant from n of the column or the row of matrix. Here,this property can be used to get the Radon moments. In the third place, the rotational invariant property of the moment matrix is used to get the rotational invariant. In this way, the character caling and rotational invariant is constructed in this algorithm. Finally, the Gaussian rmalize the different sub-characters distance to the shape feature of image. The shape terying image and other images is computed by the Euclidean distance. Experiments tethod is of robustness to the noises in image＇s similarity retrieval and has a higher hose of Pseudo-Zernike moments, wavelet modulus maximum and Tchebichef moments.

关 键 词：基于内容的图像检索 RADON变换 小波变换 奇异值 

分 类 号：TN911.73[电子电信—通信与信息系统]