DoG小波的对称性图像分类  

Classification of symmetric images by using DoG wavelet

作　　者：鲁坚[1] 邹玉茹[1] 

机构地区：[1]深圳大学数学与计算科学学院,广东深圳518060

出　　处：《计算机工程与应用》2009年第2期7-9,12,共4页Computer Engineering and Applications

基　　金：国家自然科学基金No.60873168;广东省自然科学基金No.2008257;深圳大学科研启动项目(No.200850;No.200864)~~

摘　　要：提出了一种DoG小波域的循环群、二面体群对称性图像的识别与分类技术。该技术具有很好的噪声(仅讨论加性的白高斯噪声)鲁棒性特点,甚至在高噪声情况下(例如噪声偏差σ=70),结合修改的脊波变换能实现对噪声图像包含的循环群、二面体群对称性进行识别和分类。An effective algorithm for detection and classification of cyclic and dihedral symmetric images in DoG wavelet domain is presented.The proposed algorithm is very robust to noise for the noisy images corrupted by Additive White Gaussian Noise （AWGN）.Combined with the modified ridgelet transform,the proposed method can successfully detect and classify the noisy images with cyclic and dihedral symmetries,even under the situation of high-level noise（for example,noise deviation is σ=70）.

关 键 词：循环群 二面体群 对称性分类 DoG小波变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]