Multi-information Fusion Algorithm For Face Recognition Based On Gabor Wavelet

The face recognition technology is one of the most important methods in thebiometric-based authentication technology. Automatic authentication based on facerecognition has important theoretical significance and application value, it is also a hotresearch directions in the field of computer vision. However, under the non-bindingenvironment, due to the illumination, occlusion, facial expressions, and age change factors,the recognition rate and robustness of the face recognition system is still not satisfactory.This paper introduces the commonly used face recognition algorithm, and alsointroduces the basic application and principles of the Gabor wavelet and LBP operator.Make full use of the advantages of the Gabor wavelet and local binary patterns, a morestable and better expression of the face has been proposed. The innovation of thealgorithm is that information jointly contained in image space, scale and orientationdomains can provide more fully human facial feature extraction and improve theperformance of the recognition systems.Secondly, statistical uniform pattern of LBP has been proposed based on the uniquecharacteristics for different images. Unlike the method proposed by Ojala that the vastnumber of LBP mode contains only up to twice from0-1or1-0transition, we adopt amore general strategy and define the uniform pattern in a iterative way. In each step, thepatterns corresponding to the two smallest occurrence percentage collapse into a singleone and then the histogram in resorted. Eventually reduce the dimensions of the data.Finally, we use of PCA and FLDA linear classification algorithm for classification.Experimental results show that the algorithm can effectively improve the recognition rate.