Enhancing Face Recognition for Security Systems: An Approach Using Gabor Wavelet, t-SNE, and SVM

Abstract

Facial recognition is crucial for safety and security, especially for identifying people. This paper applies facial recognition to a database of facial images by analyzing the images and subsequently assigning a set of unique features to each one. The process of extracting features from the input image is accomplished using the gabor wavelet transform. t-SNE (tdistributed Stochastic Neighbor Embedding) select and reduce the dimension of features, thus specifying various aspects within the input image. These features are then used in a classification step, where a multiclass Support Vector Machine (SVM) is employed to categorize the face. Three popular databases (Yale, ORL and JAFFE) were the sources of the images used to evaluate the effectiveness of the proposed technique. The results show the system’s high accuracy in identifying facial images. Specifically, our method achieved a 97.78% accuracy rate on the Yale, 97.50 % in the ORL databases and 100 % in the JAFFE databases, outperforming traditional methods by 2%. These results approved the system’s accuracy in recognizing facial images.