Face Recognition Based on DCT and LDA

尹洪涛,付平,沙学军
DOI: https://doi.org/10.3321/j.issn:0372-2112.2009.10.018
2009-01-01
Abstract:A face recognition method based on the discrete cosine transform and linear discriminant analysis is presented.Discrete cosine transform cannot compress the data and it only maps the resource data to another data field.How to select the DCT coefficients that are most effective to classify is an important problem.In order to obtain the effective features,a feature selection algorithm according to the separability discriminant is use to select the DCT coefficients.The experiments on ORL and Yale face databases show that the improved method is effective.
What problem does this paper attempt to address?