应用轮廓波和ICA的掌纹去噪与识别  

PALMPRINT DENOISING AND RECOGNITION WITH CONTOURLET AND ICA

作　　者：颜廷秦[1,2] 孙加存[1] 周昌雄[1] 

机构地区：[1]苏州市职业大学电子信息工程系,江苏苏州215104 [2]苏州市数字化设计与制造技术重点实验室,江苏苏州215104

出　　处：《计算机应用与软件》2014年第3期156-158,共3页Computer Applications and Software

基　　金：国家自然科学基金项目(60970058);江苏省自然科学基金项目(BK2009131);江苏省青蓝工程项目;苏州市科技基础设施建设计划项目(SZS201009);苏州市职业大学创新团队建设项目(3100125);苏州市职业大学校级课题(2012SZDYY04)

摘　　要：在掌纹识别的实际操作过程中,不可避免地会受到噪声的影响,为了增强掌纹图像信息,需要去噪。轮廓波变换能够便捷地描述自然图像中的方向和纹理信息,掌纹图像纹理信息丰富,所以应在掌纹识别中引入轮廓波去噪,以突出纹理特征,进而提高识别率。提出基于轮廓波去噪的ICA(Independent component analysis)掌纹识别算法,先对掌纹图像进行轮廓波去噪,利用ICA实现掌纹特征提取与识别。基于香港理工大学掌纹数据库的仿真结果显示,轮廓波去噪方法的识别率高于小波去噪方法,说明这种方法具有一定的理论研究意义和实用价值。During the course of palmprint recognition, it is unavoidable to be influenced by noises. In order to enhance palmprint image information, it is necessary to denoise. Contourlet can conveniently describe the direction and texture information in natural images with plenty of texture information in palmprint images, therefore contourlet is introduced into palmprint recognition to highlight texture features and further improve recognition rate. The paper proposes ICA palmprint recognition algorithm based on contourlet denoising, which firstly denoises palmprint images by contourlet, then implements palmprint feature extraction and recognition by ICA. Simulation results based on Hong Kong Polytechnic Unversity palmprint database show that the recognition rate of contourlet denoising niethod is higher than that of wavelet denoising, thus illustrating that the proposed method is of both theoretical research meaning and practical application value.

关 键 词：轮廓波 独立分量分析 掌纹识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]