TY - GEN
T1 - Improved face recognition under varying illumination conditions using webface-based local descriptors
AU - Boualleg, Abdelhalim
AU - Hakim, Doghmane
AU - Deriche, Mohamed
AU - Houcine, Bourouba
AU - Sedraoui, Moussa
AU - Azzeddine, Menasria
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - An improved face recognition-based method subjected to varying lighting is proposed. It is performed through achieving the following steps. First, Contrast Equalization (CE) is integrated into the framework of Weberface (WF) method. It reduces as much the lighting impact on the input face image as possible, so it becomes more robust than the one ensured by the standard WF method. Second, The separate implementation of the two distinct local descriptors, which are Binarized Statistical Image Features (BSIF) and Local Phase Quantization (LPQ), as well as the fusion of them descriptors allows enhancing the classification accuracy. They are chosen due to their ability to discriminate and their robustness against a variety of facial variations. The proposed methodology is assessed on two public face datasets. It is found that the identification accuracy of these two extended Yale B and AR datasets is quite challenging as compared to all existing methodologies. The simulation results demonstrate the superiority of the proposed method over existing ones in terms of efficiency and improvement ratio.
AB - An improved face recognition-based method subjected to varying lighting is proposed. It is performed through achieving the following steps. First, Contrast Equalization (CE) is integrated into the framework of Weberface (WF) method. It reduces as much the lighting impact on the input face image as possible, so it becomes more robust than the one ensured by the standard WF method. Second, The separate implementation of the two distinct local descriptors, which are Binarized Statistical Image Features (BSIF) and Local Phase Quantization (LPQ), as well as the fusion of them descriptors allows enhancing the classification accuracy. They are chosen due to their ability to discriminate and their robustness against a variety of facial variations. The proposed methodology is assessed on two public face datasets. It is found that the identification accuracy of these two extended Yale B and AR datasets is quite challenging as compared to all existing methodologies. The simulation results demonstrate the superiority of the proposed method over existing ones in terms of efficiency and improvement ratio.
KW - face identification
KW - local descriptors
KW - pattern classification varying illumination
UR - https://www.scopus.com/pages/publications/85147255828
U2 - 10.1109/STA56120.2022.10018979
DO - 10.1109/STA56120.2022.10018979
M3 - Conference contribution
AN - SCOPUS:85147255828
T3 - 2022 IEEE 21st International Conference on Sciences and Techniques of Automatic Control and Computer Engineering, STA 2022 - Proceedings
SP - 375
EP - 379
BT - 2022 IEEE 21st International Conference on Sciences and Techniques of Automatic Control and Computer Engineering, STA 2022 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 21st IEEE International Conference on Sciences and Techniques of Automatic Control and Computer Engineering, STA 2022
Y2 - 19 December 2022 through 21 December 2022
ER -