CAN THE COMBINATION OF FACIAL FEATURES ENHANCE THE PERFORMANCE OF FACE RECOGNITION?

Abstract

The field of computer vision and pattern recognition has shown great interest in facial recognition due to its wide range of applications. These applications span across historical and genealogical research, forensic science, searching for missing family members, analyzing social media, automatically managing and annotating image databases, and identifying kinship relationships. This research paper aims to address the challenges associated with facial recognition by introducing two innovative approaches: Fusion-based Classifier Combination (FCC) and Sequential CNN Deep learning-based face recognition (S-CNN). In the first part of the study, we assess the effectiveness of three techniques: Local Binary Patterns (LBP), Histogram of Oriented Gradients (HOG), and a hand-crafted learned technique called Compact Binary Facial Descriptors (CBFD). To overcome these challenges, we employ a classification step that utilizes a novel multi-classifier combination model. In the second part, we propose a novel method where we extract high-level features from multiple image regions treated as sequential data using ensemble of Convolutional Neural Networks (CNNs). These features are then fed into a Deep Neural Network (DNN) for facial recognition. The experimental results obtained from well-known face databases, including Labeled Faces in the Wild (LFW) and ORL, highlight the competitive performance of both the proposed multi-classifier combination model and the S-CNN deep learning model when compared to state-of-the-art methods .