Face Recognition Based on NMF Group Strategy

宿韬,张强,魏小鹏,周昌军
DOI: https://doi.org/10.3969/j.issn.1000-3428.2009.04.070
2009-01-01
Abstract:A method of applying Non-negative Matrix Factorization(NMF) group strategy for face recognition is put forward.It divides the training images into groups.NMF is applied to the each group’s images to obtain the non-negative subspace constructed by basic image of each group’s images,project the training images and testing images to each feature subspace.The extracted feature coefficents of each group’s images are mixed for recognition based on the nearest neighbor principle.Simulation experiments illustrate the effectivity of the method on the ORL face database.
What problem does this paper attempt to address?