边界判别投影在人脸识别上的应用
|
摘要
针对最大边界准则和边界Fisher分析在人脸特征提取过程中的不足,提出一种边界判别投影降维算法,利用类样本均值与其同类边界样本定义类内离散度,利用类样本均值与其异类边界样本定义类间离散度。同时结合最大边界准则解决类内离散度矩阵奇异的问题。与经典的最大边界准则和边界Fisher分析算法相比,可以同时考虑样本的全局结构和局部结构,避免小样本问题。在人脸数据集上的实验表明,边界判别投影是一种有效的特征提取算法,提高了人脸识别准确率。
For the weakness of maximum margin criterion and margin fisher analysis in the process of human face feature extraction, this paper presents margin discriminant projection algorithm. We define within-class scatter matrix using class samples' mean and it's marginal samples of same class, and define the between-class scatter matrix using class samples' mean and it's marginal samples of other classes. At the same time, the maximum margin criterionis used to solve singularity of within-class scatter matrix. Compared with the classical maximum margin criterion and margin fisher analysis algorithm, margin discriminant projection can consider the globaland local structure of samples at the same time, avoid small sample problem. The experiments on the face datasets show that the margin discriminant projection is a kind of effective feature extraction algorithm and has improved face recognition accuracy.
For the weakness of maximum margin criterion and margin fisher analysis in the process of human face feature extraction, this paper presents margin discriminant projection algorithm. We define within-class scatter matrix using class samples' mean and it's marginal samples of same class, and define the between-class scatter matrix using class samples' mean and it's marginal samples of other classes. At the same time, the maximum margin criterionis used to solve singularity of within-class scatter matrix. Compared with the classical maximum margin criterion and margin fisher analysis algorithm, margin discriminant projection can consider the globaland local structure of samples at the same time, avoid small sample problem. The experiments on the face datasets show that the margin discriminant projection is a kind of effective feature extraction algorithm and has improved face recognition accuracy.
引文
[1] HERRMANN F J, FRIEDLANDER M P, YILMAZ O. Fighting the curse of dimensionality: Compressive sensing in exploration seismology[J]. IEEE Signal Processing Magazine, 2012, 29(3):88-100.
[2] MOON H, PHILLIPS P J. Computational and performance aspects of PCA-based face-recognition algorithms[J]. Perception, 2001, 30(3):303-321.
[3] YU H, YANG J. A direct LDA algorithm for high-dimensional data — with application to face recognition[J]. Pattern Recognition, 2001, 34(10):2067-2070.
[4] API F R. Face recognition using Laplacianfaces[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2005, 27(3):328-340.
[5] TALWALKAR A, KUMAR S, ROWLEY H. Large-scale manifold learning[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2008:1-8.
[6] CHOI H, CHOI S. Robust kernel isomap[J]. Pattern Recognition, 2007, 40(3):853-862.
[7] BENGIO Y, VINCENT P, DELALLEAU O, et al. Out-of-sample extensions for LLE, isomap, MDS, eigenmaps, and spectral clustering[C]. International Conference on Neural Information Processing Systems, MIT Press, 2003:177-184.
[8] BELKIN M, NIYOGI P. Laplacian eigenmaps for dimensionality reduction and data representation[J]. Neural Computation, 2014, 15(6):1373-1396.
[9] SI S, TAO D, CHAN K P. Discriminative hessian eigenmaps for face recognition[C]. IEEE International Conference on Acoustics Speech and Signal Processing, 2010:5586-5589.
[10] ZHANG Z Y, ZHA H Y. Principal manifolds and nonlinear dimension reduction via local tangent space alignment[J]. Journal of Shanghai University, 2004, 8(4):406-424.
[11] 黄启宏, 刘钊. 流形学习中非线性维数约简方法概述[J]. 计算机应用研究, 2007, 24(11):19-25.
[12] CHEN Z, ZHANG B, XI W, et al. IMMC: Incremental maximum margin criterion[J]. Infection & Immunity, 2013, 82(6):2378-2389.
[13] MARONIDIS A, TEFAS A, PITAS I. Subclass marginal fisher analysis[C]. IEEE Symposium Series on Computational Intelligence, 2015:1391-1398.
[14] 甘炎灵, 金聪. 间距判别投影及其在表情识别中的应用[J]. 计算机应用, 2017, 37(5):1413-1418.
[15] HORTON P, NAKAI K. Better prediction of protein cellular localization sites with the k nearest neighbors classifier[C]. PubMed, 1997:147-152.
[2] MOON H, PHILLIPS P J. Computational and performance aspects of PCA-based face-recognition algorithms[J]. Perception, 2001, 30(3):303-321.
[3] YU H, YANG J. A direct LDA algorithm for high-dimensional data — with application to face recognition[J]. Pattern Recognition, 2001, 34(10):2067-2070.
[4] API F R. Face recognition using Laplacianfaces[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2005, 27(3):328-340.
[5] TALWALKAR A, KUMAR S, ROWLEY H. Large-scale manifold learning[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2008:1-8.
[6] CHOI H, CHOI S. Robust kernel isomap[J]. Pattern Recognition, 2007, 40(3):853-862.
[7] BENGIO Y, VINCENT P, DELALLEAU O, et al. Out-of-sample extensions for LLE, isomap, MDS, eigenmaps, and spectral clustering[C]. International Conference on Neural Information Processing Systems, MIT Press, 2003:177-184.
[8] BELKIN M, NIYOGI P. Laplacian eigenmaps for dimensionality reduction and data representation[J]. Neural Computation, 2014, 15(6):1373-1396.
[9] SI S, TAO D, CHAN K P. Discriminative hessian eigenmaps for face recognition[C]. IEEE International Conference on Acoustics Speech and Signal Processing, 2010:5586-5589.
[10] ZHANG Z Y, ZHA H Y. Principal manifolds and nonlinear dimension reduction via local tangent space alignment[J]. Journal of Shanghai University, 2004, 8(4):406-424.
[11] 黄启宏, 刘钊. 流形学习中非线性维数约简方法概述[J]. 计算机应用研究, 2007, 24(11):19-25.
[12] CHEN Z, ZHANG B, XI W, et al. IMMC: Incremental maximum margin criterion[J]. Infection & Immunity, 2013, 82(6):2378-2389.
[13] MARONIDIS A, TEFAS A, PITAS I. Subclass marginal fisher analysis[C]. IEEE Symposium Series on Computational Intelligence, 2015:1391-1398.
[14] 甘炎灵, 金聪. 间距判别投影及其在表情识别中的应用[J]. 计算机应用, 2017, 37(5):1413-1418.
[15] HORTON P, NAKAI K. Better prediction of protein cellular localization sites with the k nearest neighbors classifier[C]. PubMed, 1997:147-152.