Local Feature Based CNN for Face Recognition

Abstract

In recent years, face recognition has been a research hotspot due to its advantages for human identification. Especially with the development of CNN, face recognition has achieved a new benchmark. However, the construction of Convolutional Neural Network (CNN) requires massive training data, to alleviate the dependence on data size, a face recognition method based on the combination of Center-Symmetric Local Binary Pattern (CSLBP) and CNN is proposed in this paper. The input image of CNN is changed from the original image to the feature image obtained by CSLBP, and the original image is subjected to illumination preprocessing before the feature image is extracted. Experiments are conducted on FERET databases which contain various face images. Compared with the CNN, the method CSLBP combined with CNN that we proposed achieves the satisfying recognition rate.