Fast Local Binary Patterns for Efficient Face Recognition

Abstract

LBP, Local Binary Patterns, is an accepted technique for efficient face recognition. The local features improve the recognition process. However, high memory and computational resources are needed for LBP required approaches to improve the performance. Many people used LBP for extracting features and Support Vector Machine (SVM), histogram matching, neural networks as recognition tools. These approaches consume considerable computational resources. We have proposed a fast LBP which uses Two-level Correlation for the classification & recognition. The exhaustive experiments on FERET database 8750 images substantiate the performance compared to others. [Keywords— Face Recognition, LBP, Histogram Matching, Two-level Correlation, FERET data set]