Fuzzy Based Face Recognition System using Machine Learning Algorithms

ABSTRACT: Face recognition is a hard computer vision problem with increasingly more applications in finance, healthcare, and security. Yet, being insensitive to varying illumination, facial expressions, poses, and occlusions is still a hard problem. Current systems are non-robust and generalize poorly due to poor feature extraction and inefficient classifier performance. Experiments have to be performed with the implementation of strong feature extraction methods and high-performance classifiers to enhance system performance. Hybrid methods using the combination of various features may possess potential for accomplishing higher recognition accuracy and reliability. This paper provides a comparative analysis of ANN, SVM, and RF classifiers using HOG, fuzzy FIS, and hybrid combinations of features for face recognition. The proposed model indicates that fuzzy-based feature extraction and hybrid features improve the system's robustness, accuracy, and reliability when tested on the ORL dataset. KEYWORDS: Face Recognition, Artificial Neural Networks (ANN), Support Vector Machine (SVM), Random Forest (RF), Histogram of Oriented Gradients (HOG), Fuzzy Inference System (FIS), Hybrid Feature Extraction.