Statistical face image preprocessing and non-statistical face representation for practical face recognition

Recognizing face images in real environment is still an challenging problem since there are severe illumination changes. In this paper, we propose a practical face recognition method that combines statistical global illumination transformation and non-statistical local face representation method. When a new face image is given, it is transformed into a number of face images exhibiting different illuminations using a statistical bilinear model-based indirect illumination transformation. Each illumination transformed image is then represented by a histogram sequence that concatenates the histograms of the non-statistical multi-resolution uniform local Gabor binary patterns (MULGBP) for all the local regions. To facilitate this, the input image is divided into several regular local regions, each local region is converted into several Gabor filters, and each Gabor filtered region image is converted into multi-resolution local binary patterns (MULBP). Finally, face recognition is performed by a simple histogram matching process. Experimental results show that proposed face recognition method is highly robust to illumination variation as exhibited in the real environment.