Research And Implementation Of Multimode Fusion Identity Recognition System For Mobile Payment

With the continuous development of communication networks and smart phones,mobile payment has become one of the mainstream ways of transactions in people's daily life,making people's lives more convenient.At present,Alipay and We Chat have added face recognition payment and fingerprint identification payment to the single mode certification payment mode,which has enhanced the payment environment.However,there are still some drawbacks in single-mode feature recognition such as face and fingerprint.For example,the recognition effect of single-mode biological features in complex environments such as illumination and expression transformation will be affected,and it is also easy to be forged by illegal molecules.In order to solve these drawbacks and further improve the security performance and recognition performance,a very potential pattern,multi-mode fusion recognition,emerged as the times require.In this paper,face and fingerprint are used to fuse the two biological features.Firstly,face and fingerprint are preprocessed to optimize the quality of the original image to ensure better recognition effect.Face image preprocessing is relatively simple,the focus is on the preprocessing of fingerprint image.The collected fingerprint image is disturbed by many factors,which will seriously affect the fingerprint recognition performance.This paper calculates the gradient field and direction field of the fingerprint image,segmentes the fingerprint image,enhances the fingerprint image,binarizes the fingerprint image and thins the fingerprint image,and finally obtains the high quality fingerprint image.Secondly,feature extraction and classification of pre-processed face images and fingerprint images are carried out.In face recognition,in order to solve the disadvantage that the recognition effect of single-mode biological features will be affected in complex environments such as illumination and expression transformation,this paper proposes a face recognition algorithm based on sparse representation,and uses PCA dimensionality reduction algorithm to extract features to reduce computational complexity.In fingerprint recognition,this paper adopts the feature extraction algorithm based on thinning image,and applies the algorithm of removing false feature points to ensure the accuracy of the extracted feature points.Finally,the matching level is chosen to fuse.Compared with other levels of fusion,the matching level is simple to implement and fast to calculate while guaranteeing the recognition effect.In this paper,we propose a G-WMF matching layer fusion algorithm for face fingerprint.After getting the matching score of face and fingerprint,we normalize it,then fuse the weight mean,and add genetic and evolutionary algorithms to assign the best weight to face and fingerprint at a relatively fast speed.Combining with the face fingerprint fusion algorithm,this paper also designs a multi-mode fusion identity recognition system for mobile payment.In the aspect of demand analysis and function analysis,this paper discusses it.