ICCV-MoG
所属分类：matlab编程
开发工具：matlab
文件大小：2309KB
下载次数：47
上传日期：2017-09-14 11:08:28
上 传 者：
tiantian21
说明： 孟德宇老师的低秩分解代码，对动态变化的视频帧提取它的低秩背景，同时得到它的稀疏部分。利用混合高斯对误差进行建模
(Teacher Meng Deyu's low rank decomposition code extracts the low rank background of the dynamic video frame and obtains the sparse part of it. The error is modeled by mixed Gauss)
文件列表:
ICCV-MoG\DEMO_ExpClean.m (2711, 2014-12-07)
ICCV-MoG\DEMO_ExpGaussian.m (2821, 2014-12-07)
ICCV-MoG\DEMO_ExpMixture.m (3090, 2014-12-07)
ICCV-MoG\DEMO_ExpSparse.m (2835, 2014-12-07)
ICCV-MoG\LogLikelihood.m (466, 2014-02-22)
ICCV-MoG\logsumexp.m (499, 2014-02-22)
ICCV-MoG\MLGMDN.m (8988, 2014-12-07)
ICCV-MoG\Weighted L2 MF\ALM\ALM.m (3608, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_block_matrix_solve.m (2865, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_merge_matrices.m (197, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd.m (11014, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_aanaes.m (2397, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_alternation.m (1314, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_alt_dn_hybrid_v01.m (1593, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_alt_dn_hybrid_v01a.m (1711, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_alt_dn_hybrid_v02.m (2187, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_alt_dn_hybrid_v02a.m (2284, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_alt_dn_hybrid_v03.m (4372, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_brandt.m (4440, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_damped_newton.m (2309, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_dn_altreg.m (2651, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_dn_line_search.m (2227, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_dn_ls_altreg.m (2438, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_errfunc_line_search.m (5483, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_errfunc_line_search_debug.m (10117, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_errfunc_regularizeAB.m (9680, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_errfunc_sumsqrderr.m (11616, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_errfunc_sumsqrderr_gradient_mex.cxx (2912, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_errfunc_sumsqrderr_gradient_mex.mexw64 (9728, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_errfunc_sumsqrderr_hessian_mex.cxx (4317, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_errfunc_sumsqrderr_hessian_mex.mexw64 (10240, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_gradient_descent.m (1517, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_huynh.m (3967, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_matrix_dimension_check.m (1129, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_powerfactorization.m (1506, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_project_and_merge.m (1243, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_mfwmd_shum.m (1819, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\amb_parse_arguments.m (2294, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton\Untitled8.m (40, 2014-02-22)
ICCV-MoG\Weighted L2 MF\Damped Newton.pdf (356780, 2014-02-22)
... ...
Usage of damped_wiberg:
This MATLAB function factorizes a given matrix Y into U*V', a product
of two smaller matrices of rank r. To be specific, it minimizes ||H
\odot (Y - U*V')||_2 for specified data Y, H, and r, where Y is the
matrix to be factorized, H is the binary matrix indicating existing
(1) and missing (0) components of Y, and r is the number of columns of
U and V. For more detail, see the source code.
For the details of the implementation, please see the following
reference. Please cite it in your work when you use our code.
[1] Takayuki Okatani, Takahiro Yoshida, Koichiro Deguchi: Efficient
algorithm for low-rank matrix factorization with missing components
and performance comparison of latest algorithms. Proc, ICCV 2011:
842-849.
Example)
1. Load your data. (These data are from
http://www.robots.ox.ac.uk/~abm/. Please download the original when
you use them for your study.)
>> load('M.txt');
>> load('mask.txt');
2. Factorize the matrix.
>> [U,V]=damped_wiberg(M, mask, 4);
Done. The algorithm should ALWAYS converge to the global minimum for
about 50-300 iterations. The results are such that M <-> U*V'.
近期下载者：
相关文件：
收藏者：