Face recognition in uncontrolled conditions

Face recognition has many challenges in real conditions such as pose variation, low resolution, facial expression, illumination changes, blurriness and occlusion. This paper introduces a novel face recognition system which works under these real conditions. The system uses multi scale landmark-based histogram of gradient for feature extraction and PCA whitened space for classification. To show the efficiency of our system, it is tested on the FERET database in addition to the database collected by this paper. The collected database is provided in real conditions including all the mentioned challenges.