基于稀疏子空间聚类的人脸识别方法
|
摘要
在现有的稀疏子空间聚类算法理论基础上给出两种稀疏子空间聚类优化算法:稀疏线性子空间聚类和稀疏仿射子空间聚类。这两种优化算法针对不同的数据集会有不同的聚类效果。通过稀疏表达得到不同的稀疏系数矩阵,把稀疏系数矩阵应用到较为简单的改进的正则化谱聚类算法中实现聚类。应用Yale B数据对人脸图像进行识别分类得出:采用稀疏线性子空间聚类算法优于稀疏仿射子空间聚类算法;在算法执行时间上和算法聚类错误率比传统的稀疏子空间聚类较为快速高效。
We offer two kinds of sparse subspace clustering optimization algorithm,sparse linear space clustering and sparse affine subspace clustering,based on the existing theory of sparse subspace clustering algorithm. For different data gathering,these two kinds of optimization algorithm has different clustering results. In this paper,different sparse coefficient matrix by sparse expression is obtained. In order to achieve cluster,the sparse coefficient matrix is applied to relatively simple regularization of spectral clustering algorithm. Application of Yale B data,we recognize and classify face image :using sparse linear space clustering algorithm is better than the sparse affine subspace clustering algorithm;Comparing with the traditional sparse subspace clustering,it is more fast and efficient in the time of execution and error rate of algorithm.
We offer two kinds of sparse subspace clustering optimization algorithm,sparse linear space clustering and sparse affine subspace clustering,based on the existing theory of sparse subspace clustering algorithm. For different data gathering,these two kinds of optimization algorithm has different clustering results. In this paper,different sparse coefficient matrix by sparse expression is obtained. In order to achieve cluster,the sparse coefficient matrix is applied to relatively simple regularization of spectral clustering algorithm. Application of Yale B data,we recognize and classify face image :using sparse linear space clustering algorithm is better than the sparse affine subspace clustering algorithm;Comparing with the traditional sparse subspace clustering,it is more fast and efficient in the time of execution and error rate of algorithm.
引文
[1]TSENG P.Subspace clustering[J].Signal Processing Magazine,2001,28(2):52-68.
[2]HO J,YANG M H,LIM J,et al.Clustering appearances of objects under varying illumination conditions[C]//IEEE Conference on Computer Vision and Recognition,2003.
[3]COSTERIRA J,KANADE T.A multibody factorization method for independently moving objects[J].International Journal of Computer Vision,1998,29(3):159-179.
[4]TIPPING M,BISHOP C.Mixtures of probabilistic principal component analyzers[J].Neural Computation,1999,11(2):443-482.
[5]WEISS Y.On spectral clustering:analysis and algorithm[J].Neural Information Processing Systems,2001:849-856.
[6]YAN J,POLLEFEYS M.A general framework for motion segmentation:Independent.Articulated,rigid.non-rigid.Degenerate and non-degenerate[C]//in European Conf.on computer,2006:94-106.
[7]CHEN G,LERMAN G.Spectral curvature clustering(SSC)[J].International Journal of Computer Vision,2009,81(3):317-330.
[8]ELHAMIFAR E,VIDAL R.Sparse subspace clustering:Algorithm,theory,and applications[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2013,35(11):2765-2781.
[9]BOYD S,PARIKH N,CHU E,et al.Distributed optimization and statistical learning via the alternating direction method multipliers[J].Foundations and Trends in Machine Learning,2010,3(1):1-122.
[10]LUXBURG U.A tutorial on spectral clustering[J].Statistics and Computing,2007,17(4):395-416.
[11]欧阳佩佩,赵志刚,刘桂峰.一种改进的稀疏子空间聚类算法[J].青岛大学学报(自然科学版),2014,27(3):44-48.
[12]苏木亚.谱聚类方法研究及其在金融时间序列数据挖掘中的应用[D].大连:大连理工大学,2011.
[13]封睿,李小霞.基于GLC KSVD的稀疏表示人脸识别算法[J].四川兵工学报,2014,35(4):88-92.
[2]HO J,YANG M H,LIM J,et al.Clustering appearances of objects under varying illumination conditions[C]//IEEE Conference on Computer Vision and Recognition,2003.
[3]COSTERIRA J,KANADE T.A multibody factorization method for independently moving objects[J].International Journal of Computer Vision,1998,29(3):159-179.
[4]TIPPING M,BISHOP C.Mixtures of probabilistic principal component analyzers[J].Neural Computation,1999,11(2):443-482.
[5]WEISS Y.On spectral clustering:analysis and algorithm[J].Neural Information Processing Systems,2001:849-856.
[6]YAN J,POLLEFEYS M.A general framework for motion segmentation:Independent.Articulated,rigid.non-rigid.Degenerate and non-degenerate[C]//in European Conf.on computer,2006:94-106.
[7]CHEN G,LERMAN G.Spectral curvature clustering(SSC)[J].International Journal of Computer Vision,2009,81(3):317-330.
[8]ELHAMIFAR E,VIDAL R.Sparse subspace clustering:Algorithm,theory,and applications[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2013,35(11):2765-2781.
[9]BOYD S,PARIKH N,CHU E,et al.Distributed optimization and statistical learning via the alternating direction method multipliers[J].Foundations and Trends in Machine Learning,2010,3(1):1-122.
[10]LUXBURG U.A tutorial on spectral clustering[J].Statistics and Computing,2007,17(4):395-416.
[11]欧阳佩佩,赵志刚,刘桂峰.一种改进的稀疏子空间聚类算法[J].青岛大学学报(自然科学版),2014,27(3):44-48.
[12]苏木亚.谱聚类方法研究及其在金融时间序列数据挖掘中的应用[D].大连:大连理工大学,2011.
[13]封睿,李小霞.基于GLC KSVD的稀疏表示人脸识别算法[J].四川兵工学报,2014,35(4):88-92.