SSIM Algorithm In3D Face Recognition

With the high development of the information society, people pay more and more attention to systems security and information security. As a mainstream method of the biometric recognition,3D face recognition combines with mathematics and advanced technology. It has been an important mean for identity authentication because of its user friendly interface and invariance to illumination, pose variations and makeup.But, the existing recognition methods are not good for wholly improving face recognition speed because of one-time complexity recognition. This paper uses novel methods which are angle-based SSIM and normal vector-based SSIM to face recognition step by step.Firstly, statistical methods are applied to analyze the stability of face areas and locate the stable regions where are nose region and forehead region. Secondly, seven representative curves are extracted from face surface through cubic spline interpolation method. After normalization, these depth curves are changed to angel curves for weakening the impact of gesture. Similarity is calculated with SSIM by score fusion to initial recognition. Secondly, normal vectors which belong to nose region and forehead region are got through least squares surface fit, then, cosine values of angles between3axes and normal vectors are calculated. Cosine values which are on nose edge and forehead surface are obtained by making use of image processing and statistical method to compose characteristics. After that, similarity score is acquired by computing characteristics’similarity with SSIM and score fusion for last recognition.This paper’s experiment makes use of part of FRGC2.0dataset, which doesn’t have expression or has a little expression, to test the stability of the algorithm. The experimental results show that this method can enhance the average speed of the whole system.