Face Recognition And Facial Expression Recognition Based On Local Binary Patterns

With the recovery of artificial intelligence, the international community has set off a boom in the development of artificial intelligence. Face recognition and facial expression recognition as two important branches of intelligent recognition, has been the focus of research scientists. Face recognition and fingerprint recognition, iris recognition, gait recognition, palm recognition belonging to biometrics recognition, involves many disciplines such as computer vision physiology with, due to its intuitive, non-contact, a user can easily accept the other notable features gradually become scientist’s hot research and has been successfully used in security systems. However, due to the complexity and universal recognition technology environment itself is defective, face recognition system is not strong, is still facing a lot of problems to be solved. In recent years, facial expression recognition technology as computer science and human-computer interaction emerging research topics obtains more and more attention. Facial recognition technology is a cross-discipline involving physiology, psychology, machine vision, and many other research areas, with a wide range of application value.Local Binary Pattern (local binary patterns, LBP) is a texture description operator, because the local binary pattern operator calculation is very simple features after filtering LBP operator obtained has a strong ability to distinguish, which makes a lot of researchers introduce it into different application areas and achieved good results. This article will apply local binary pattern into face recognition and facial expression recognition. Local Binary Pattern for feature extraction operator is not completed, this paper completed the following tasks:(1) Propose a new improved model of local binary operator (Merge Local Binary Patterns, MLBP). The proposed operator is more complete than the original local binary pattern feature extraction operator model, and more robust for illumination changes and occlusion. Experiments in the AR face database, CMU-PIE face database show that this operator is more effective and superior.(2) For a low face recognition rate using local binary pattern operator when the light changes, this paper proposed combining the wavelet transform with local binary pattern, wavelet transform uneven illumination face image processing, retention less affected by the high-frequency portion of the light, and then use the local binary pattern feature extraction. This algorithm reduces the impact of a good light on the recognition rate, greatly improving the recognition rate when uneven illumination experiment at CMU-PIE database and YALEB library prove the feasibility of the algorithm.(3) For facial expression recognition paying more attention to the details of expression, this paper proposes a local binary operator (Divided Local Binary Patterns, DLBP) based on parity mode decomposition, the operator continue to break down on the basis of the traditional LBP operator, the original LBP operator is sub-divided into two LBP operators in accordance with the idea of the Fourier decomposition of parity. Improved DLBP operator not only greatly reduces the feature dimension, also extract details of the expression better. Experiments in Japanese women face library explained a high recognition rate with DLBP operator.This paper studies applications of the local binary patterns in face recognition and facial expression recognition of the different needs of occasions when the original LBP operator to do the different improvements made different improvements LBP operator, to promote the automatic Face recognition systems and the further development and application of automatic facial expression recognition system has a positive meaning.