A Two-Phase Framework based on 2-D feature extraction algorithms

In this paper, we proposed a two-phase framework based 2-D feature extraction algorithms to overcome the common disadvantage which 2-D algorithms have. To verify the validity of the framework, a series of experiments were performed on the ORL database based on three 2-D algorithms: 2DPCA, 2DLDA and 2DLEM. The results show that the 2-P framework can reduce the coefficients effectively and have achieved the approximate performance to direct 2-D algorithms. The experiments have also indicated that 2DLEM has outstanding performance not only in the direct 2-D algorithms but in the 2-P framework