Local Binary Pattern and Principal Component Analysis for Low-light Face Recognition

Facial recognition technology allows both the government and the general public to identify individuals based on their facial features, even in cases of significant changes. However, low lighting during the facial recognition process can lead to confusion during feature extraction, thus affecting facial recognition performance. This research aims to identify the most effective method for recognizing faces in low-light conditions. The Local Binary Pattern (LBP) and Principal Component Analysis (PCA) algorithms each have their strengths in feature extraction. These algorithms will be combined with others, including Histogram of Oriented Gradient (HOG), K-Nearest Neighbor (KNN), and Support Vector Machine (SVM). The process begins with face detection, followed by feature extraction using LBP, PCA, and HOG. Subsequently, these features will be classified using KNN and SVM. The combinations of these algorithms tested in this research include LBP-KNN, LBP-HOG-PCA-SVM, HOG-PCA-SVM, and MobileNet for additional comparison. The results indicate that the HOG-PCA-SVM combination achieved the highest accuracy among the other combinations, reaching 98%. This makes the HOG-PCA-SVM combination a preferred choice for low-light face recognition systems without image enhancement.