Method Study for Facial Character Recognition Based on Improved LBP

摘要

Facial character recognition is very important technology in the security control system, correspondingly, and also it is widely used for lots of application fields. Aimed at some puzzles of ignoring the pixel contrast values of local region, discarding some important texture characteristic information and so on in existing traditional facial character recognition algorithms, the paper proposed an improved LBP method for facial character recognition under complex illumination. The method firstly made uniformization pretreatment for different illumination facial image, limited the illumination change to a certain extent, then mapped the pixel contrast value of local region to an interval value so as to make the fluctuation be controlled in an interval, and recognized facial character finally. It took the face recognition as an example to make the simulation, and the experiment simulation results based on CMU PIE, Yale B and Outdoor facial Database demonstrated that the ILBP could be better than traditional facial character algorithms. Research results show that the proposed method owns better recognition effect, and therefore it is more reasonable and effective.