Face recognition using fusion of feature learning techniques

• Published in: Measurement : journal of the International Measurement Confederation Vol. 146; pp. 43 - 54
• Main Authors: Umer, Saiyed, Dhara, Bibhas Chandra, Chanda, Bhabatosh
• Format: Journal Article
• Language: English
• Published: London Elsevier Ltd 01.11.2019; Elsevier Science Ltd
• Subjects: Classification; Face recognition; Facial recognition technology; Feature extraction; Feature learning; Feature recognition; Frontal; Fusion; Learning; Profile; Representations; Studies; Support vector machines; Feature learning; Frontal; Face recognition; Fusion; Profile
• ISSN: 0263-2241; 1873-412X
• DOI: 10.1016/j.measurement.2019.06.008

•A novel face recognition system has been proposed for frontal and profile faces which need the minimum computer configurations.•Parameters of the tree-structured part model have been tuned to obtain the facial landmark points for extracting the face region.•The local dense SIFT features with SRC, CD, BCD and LLC feature learning techniques followed by SPM have been employed for feature computation.•The performance of the proposed system due to feature representation schemes have been fused using post-classification fusion techniques.•The outstanding performance has been obtained for the ORL, IITK, CVL, CASIA-V5, FERET, and CAS-PEAL challenging face databases with state-of-the-art methods. A method for face recognition system for both challenging frontal and profile faces is proposed in this paper. The proposed system consists of face pre-processing, feature extraction and classification components. During pre-processing, a region-of-interest for face region is extracted based on facial landmark points, obtained by a Tree Structured Part Model. During feature extraction, Scale Invariant Feature Transform descriptors are computed from patches over detected face region. These descriptors undergo to different feature learning techniques to obtain different feature representations for the input image. The performance of these feature representations are obtained using multi-class linear Support Vector Machine classifier during classification. Finally, the scores from different feature learning techniques are fused to take the decision to recognize the subjects. Extensive experimental results have been demonstrated to show the effectiveness of the proposed face recognition system. The comparison with the exiting state-of-art methods for ORL, IITK, CVL, AR, CASIA-Face-V5, FERET and CAS-PEAL face databases, show the superiority of the proposed system.