刘玉珍,赵娜,李新春,林森.基于混合滤波LBP和PCA的掌纹识别[J].测控技术,2018,37(2):11-15 |
基于混合滤波LBP和PCA的掌纹识别 |
Palmprint Recognition Based on Hybrid Filtering LBP and PCA |
|
DOI: |
中文关键词: 局部二值模式 HFLBP 主成分分析 鲁棒性 掌纹识别 |
英文关键词:local binary pattern HFLBP principal component analysis robustness palmprint recognition |
基金项目:辽宁省教育厅科学研究一般项目(L2014132);辽宁省自然科学基金面上项目(2015020100) |
|
摘要点击次数: 1435 |
全文下载次数: 1076 |
中文摘要: |
针对基于局部二值模式(LBP)的掌纹识别易受噪声影响,导致算法鲁棒性下降,提出一种结合混合滤波LBP(HFLBP)和主成分分析(PCA)的特征提取方法。从滤波和特征提取的角度对传统LBP算法进行改进,先对图像进行去噪处理,然后对掌纹图像进行分块,提取LBP直方图特征向量,并通过PCA算法对特征向量进行降维,最后利用欧氏距离匹配。在香港理工大学PolyU图库和PolyU噪声图库上与几种典型算法进行对比实验,实验结果表明,本文算法分别获得最低等误率为1.1405%、4.0101%,有效地提高了识别率和鲁棒性,具有很好的应用前景。 |
英文摘要: |
In order to solve the problem that palmprint identification based on local binary pattern(LBP) is sensitive to noise and has poor robustness,a new feature extraction method based on hybrid filtering local binary pattern(HFLBP) and principal component analysis(PCA) is proposed.The traditional LBP algorithm is improved from the perspective of filtering and feature extraction.Firstly,the image is denoised,then the palmprint image is segmeated,the LBP histogram feature vector is extracted and reduced by PCA algorithm.Finally,the palmprint classification is implemented by Euclidean distance.Several typical methods were compared in the two databases of PolyU library and PolyU noise library of HongKong Polytechnic University.The results show that the EER obtained by this method is the lowest,which is 1.1405% and 4.0101% respectively.In addition,the algorithm enhances the recognition rate and robustness of palmprint,which has a good application prospect. |
查看全文 查看/发表评论 下载PDF阅读器 |
关闭 |