HOG and LBP: Towards a robust face recognition system

Abstract

Face recognition has been a long standing problem in computer vision. Histograms of Oriented Gradients (HOGs) and Local Binary Patterns (LBPs) have proven to be an effective descriptor for object recognition in general and face recognition in particular. In this paper, we investigate a simple but powerful approach to make robust use of HOG and LBP features for face recognition. The three main contributions of this work are: First, in order to compensate for errors in facial feature detection due to occlusions, pose and illumination changes, we propose to extract HOG descriptors from a regular grid. Second, fusion of HOG descriptors at different scales with the LBP ones allows to capture important structure for face recognition. Third, we identify the necessity of performing feature selection to remove redundant and irrelevant features to make the classification process less prone to overfitting. This is particularly important if HOG features are extracted from overlapping cells. It was observed in the course of experiments that the obtained results on three databases prove the benefits of our approach.