Occluded Face Recognition Using Feature Selection Approach

Face recognition offers significant advantages over other biometric systems, such as signatures, iris scanning, and fingerprints, due to its non-intrusive nature. However, variations in illumination, partial occlusions, and differing imaging conditions pose challenges to effective face recognition. This study presents the design, implementation, and validation of a face recognition system capable of operating under partial occlusion conditions. The process encompasses image acquisition, facial component detection, feature extraction, recognition, and identification of individuals. Distinct detectors are utilized for recognizing facial components such as the left eye, right eye, nose, and mouth. Post-detection, operations like histogram equalization, resizing, and vectorization are applied. The proposed technique incorporates a feature selection approach leveraging Principal Component Analysis (PCA), selecting 1033 features for recognition from an initial set of 2520. The FeedForward Neural Network is employed for classification. The system's efficacy is demonstrated using the AR face database, within a MATLAB environment.