Face Recognition Based On Illumination Preprocessing And Feature Fusion

Face recognition is applied to our work and daily life, and has attracted much attention in many areas, such as information security, human-computer interaction. After decades of studying face recognition, the research on face recognition has achieved a lot of fruitful results, and a large number of face recognition algorithms have been developed. But in real world, there are still many difficults to overcome. Among them, illumination is one of the most important factors that affect the performance of the face recognition. Sometimes, the change of different illuminations for the same person has greater effect than the difference among different persons with the same illumination.In order to eliminate the influence of illumination on face recognizing, a weight method of image based on Gamma correction and SQI is adapted to the source face image in the stage of image preprocessing. This weight image combines the advantages of Gamma correction and SQI. On one hand, the weight image can elimincate the influence of illumination on the image, on the other hand, the loss of texture detail in SQI is compensated. Moreover, the cell-based muti-blocks local ternary pattern, an illumination invariant feature, is presented, which is based on the LBP. This feature not only inherits the property of illumination inverance in the LBP, but improves the robustness to the noise in the face images. As a local feature, the cell-based muti-blocks local ternary pattern combines with the PCA, which is a globle feature, to decrease the dimensionality. Finally, the paper shows the whole procedure of face recognition and results of the experiments. The results show that the method presented in this paper achieves a higher accuracy compared with other related methods.