Enhancing Face Recognition in Surveillance Systems Using Local Binary Pattern and (PCA) Based Feature Selection

Abstract

Face recognition is a critical component of biometric systems utilized for surveillance to identify criminals, suspected terrorists, and missing children. This paper presents the application of the Local Binary Pattern (LBP) method for feature extraction in face recognition, recognized for its robustness and effectiveness. The primary contributions of this research include the use of LBP to extract key facial features and the implementation of Principal Component Analysis (PCA) to refine these features by eliminating irrelevant data, thereby enhancing classification accuracy. Given the high dimensionality of facial data, selecting significant features is crucial for effective recognition. For classification, two algorithms Support Vector Machine and Linear Discriminate (LD) were employed to analyze the feature vectors derived from the LBP method. Experimental validation was performed on the standard benchmark dataset from the Olivetti Research Laboratory (ORL). Such accuracy shows that the proposed system is more accurate than the previous models. I have also conducted an analysis of classifier performance with all features in comparison to classifiers refined by PCA, and though full-feature classifiers outperform PCA classifiers overall, PCA classifiers reward us with time efficiency advantages. It finds the durability and effectiveness of the proposed face recognition system with combination of LBP and PCA features