Face Recognition Model based on the Laplacian Pyramid Fusion Technique

In this paper, a Fusion Feature-Level Face Recognition Model (FFLFRM) based on the Laplacian Pyramid (LP) fusion technique is proposed. The proposed FFLFRM model consists of four main processes: face detection, feature extraction, feature fusion, and face classification. In the FFLFRM model, the important characteristics of the face (i.e., the mouth, nose, and eyes) are detected, as well as, both global and local features are extracted using Principal Component Analysis (PCA) and the Local Binary Pattern (LBP) extraction methods. The extracted features are then fused using the LP fusion technique and classified using the Artificial Neural Network (ANN) classifier. The FFLFRM model was tested on 10,000 face images generated from the Olivetti Research Laboratory (ORL) database. The performance of the FFLFRM was compared with three state-of-the-art face recognition models based on local, global, and Frequency Partition (FP) fusion techniques, in terms of illumination, pose, expression, occlusion, and low image resolution challenges. The recognition results of the proposed FFLFRM were promising. Hence, it achieved up to 98.2 recognition accuracy. Thus, shows the effectiveness of the proposed model in manipulating with variant face challenges.