Face Recognition Based On Fusion Feature Extraction Algorithms

Biometric-based methods have emerged as a highly promising avenue for individual recognition, with face recognition standing out as a key application in image analysis. Facial recognition systems are affected by various image condition factors, including expressions, occlusion, poses, and illuminations. It is crucial to consider reasonable variations in illumination between gallery and probe images when developing face recognition algorithms. In the context of face verification, illumination variation significantly influences the likelihood of misclassification. Existing research lacks a focus on imaging conditions, especially regarding illumination, in face recognition systems. This research aims to identify and design a scheme to address illumination variation in facial recognition. The proposed scheme will undergo analysis and evaluation to assess its accuracy and effectiveness. The image processing in this research is divided into two phases. The first is the preprocessing phase, encompassing auto contrast balancing and noise reduction. The second phase is face processing, which involves illumination normalization, face detection, feature extraction, and face recognition. Two datasets are utilized: the first being the ORL dataset, which includes images with illumination variations, and the second being the Yale database, where images contain various noises.