基于脊波变换和SVM的MSTAR图像分类  

MSTAR Image Classification Based on Ridgelet Transform and SVM

作　　者：李含光[1] 吴小季[1] 

机构地区：[1]南京信息工程大学计算机与软件学院,南京210044

出　　处：《武汉理工大学学报》2010年第16期60-63,88,共5页Journal of Wuhan University of Technology

摘　　要：MSTAR图像分类是当今图像研究的一个重要领域。由于图像目标的边缘与轮廓是方向性很强的奇异性所在,脊波变换能够很好地处理具有线状奇异性的图像,而且能有效的提取MSTAR图像的特征。SVM分类器在少量训练样本的情况下,可以给出较为准确的分类效果。提出了一种基于脊波变换和支持向量(SVM)的MSTAR图像分类方法,并与同类的方法进行了分析比较。The MSTAR image classification is currently an important field of image researching.As the edges and contours of target in the image are positions where highly directional singularities existed,and the ridgelet transform can well handle the images with linear singularities,so we can extract the feature of the MSTAR images efficiently.SVM classifier can give exact results with few training samples.In this paper,we put forward a method based on ridgelet transform and support vector machine for MSTAR images classification,and make comparison with other similar methods.

关 键 词：脊波变换 SVM 图像分类 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]