An Effective Method for Face Recognition by Creating Virtual Training Samples Based on Pixel Processing

As one of the most attractive branch of pattern recognition and computer vision, face recognition is becoming increasingly hot among the researchers over the past few years. However, in practical applications, the change of illumination and the limited number of available training samples always greatly affect the accuracy of face recognition. In this paper, we proposed a simple but efficient method to create new virtual samples based on pixel processing for removing the influence of illumination variation and enriching training samples size. Meanwhile, we use a representation based classification (RBC) method to perform face recognition and a weighted fusion approach is utilized to integrate representation residuals for face recognition. The experimental results show that the proposed method outperforms state-of-the-art face recognition methods.