Abstract

One of the most difficult and intriguing areas of computer vision is Face Recognition (FR). Due to the low generalization ability, the success rate of FR is affected by illumination, posture shift, and other factors. FR algorithms typically aim to solve the two problems called subject identification and verification. However, such methods have limitations because they frequently call for computer vision specialists to create useful features. Thus, this paper presents the novel framework of FR using an ensemble learning model. Initially, the face images are gathered from the datasets, which is followed by the pre-processing of the images. Once the pre-processed image is obtained, the significant features are extracted by using Local Binary Pattern (LBP) and Discrete Wavelet Transform (DWT). Then, the dimension reduction of the resultant feature is done by utilizing the Principal Component Analysis (PCA). Finally, an Optimal Ensemble Classifier (OEC) is developed that includes Support Vector Machine (SVM), Neural Network (NN), and Adaboost, where the hyperparameters are tuned optimally by Enhanced Hybrid Leader Based Optimization (EHLBO) algorithm.   The performance is validated via different parameters and compared over existing approaches. Thus, the findings ensure that it attains impressive results for face recognition.

Statistics