Post-processing of dimensionality reduction methods for face recognition

Application of MEEMD in post‐processing of dimensionality reduction methods for face recognition

“…Fig. 8 Face-identification results for SPCA, SLDA, and FNLDA with the post-processing method of Abbad et al [27] and with the OPP framework on [42] 49.00 ± 2.9 -Shen et al [24] 72. 40…”

“…The proposed method depends on a set of parameters in each phase. In this part, we will present the parameters used in our framework as well as the parameters of Zuo et al [25] and Abbad et al [27], which is the closest published work to our manuscript.…”

“…In this section, we will evaluate the performance of the proposed approach in case of grey level. The proposed system is compared with the face-recognition approach proposed by Zuo et al [25] and Abbad et al [27]. In order to test the robustness of our method for different types of dimension reduction methods, we have chosen five recent methods: sparse PCA (SPCA) [45], sparse LDA (SLDA) [46], fast null LDA (FNLDA) [47], manifold partition discriminant analysis (MPDA) [48], and discriminative transfer subspace learning via low-rank and sparse representation (TSL-LRSR) [9].…”

“…(ii) This post‐processing is not generalised; it is designed just for LDA and their variants method. Our previous work [27] attempted to overcome those drawbacks using a novel post‐processing method but, unfortunately, other problems have appeared, in that it is not discriminating enough to process images acquired in real cases and, also, it consumes a lot of time.…”

“…To address the limitations reported in [25–27], this paper proposes new discriminative post‐processing of dimensionality reduction methods. Our contributions can be summarised as follows: (i) An original adaptation of the multi‐dimensional ensemble empirical mode decomposition (MEEMD) method for the case of face recognition. (ii) In the ideal case, characteristics of images belonging to the same class should have the same dominant pattern.…”