正在进行安全检测...
发布时间:1714356846 来源:文档文库
小
中
大
字号:
第32卷第9期 2015年9月 计算机应用与软件 Computer Applications and Software Vol_32 No.9 Sep.2015 基于图像分块和特征选择的单训练样本人脸识别 李俊霞 张书敏 吴何胜 (河南农业职业学院电子信息工程系(南京大学软件学院河南郑州451450) 江苏南京210093) 摘要 为了提高人脸识别的正确率,针对单样本人脸识别训练样本存在的缺陷,提出一种基于图像分块和特征选择的单样本人 脸识别算法。首先将人脸图像划分成子块,并分别提取各子块的特征,连接成人脸图像特征向量,然后采用多流形判断分析算法选 择对人脸识别结果贡献较大的特征。最后计算采用支持向量机对人脸进行识别,并采用Yale B和PIE人脸库对本文人脸算法的有 效性和优越性进行仿真测试。仿真结果表明,相对于当前典型人脸识别算法,该算法提高了人脸识别正确率,获得了更加理想的人 脸识别效果。 关键词 训练样本人脸识别 图像分块 特征提取 中图分类号TP391 文献标识码A DOI:10.3969/j.issn.1000-386x.2015.09.073 SINGLE TRAINING SAMPLE FACE RECoGNITIoN BASED oN IMAGE BLoCKING AND FEATURE SELECTIoN Li Junxia Zhang Shumin Wu Hesheng (Department ofElectronic Information Engineering,Henan Vocational College ofAgricutM ,Zhengzhou 451450,He 口n,ChinⅡ) (Institute ofSotfware,Nanjing University,Nanjing 210093,Jiangsu,China) Abstract In order to improve the accuracy rate of face recognition,we propose a sin ̄e training sample face recognition algorithm which is based on image blocking and feature selection aimed at the defect of sin ̄e face recognition training sample.Firstly,it divides the image into sub—blocks,and extracts the features of each sub—block separately to join them to eigenvector of face image,then it Uses muhi—manifold judgement and analysis algorithm to choose the features with ̄eater contirbution to face recognition resul