Study On Face Recognition Algorithm Based On Sparse Representation

With the rapid development of information,automation and intelligence in society,the security of information is becoming increasingly important.With the proposal of "Three types,two networks,and world-class" strategic goals in the State Grid Corporation of China,it has become a trend to build a hub-type,platform-type,and shared-type enterprise,and build and operate a strong smart grid.Therefore,information security in the process of power production,management,and operation has become an important issue for power companies to resolve.The identification of staff and customers is an important part of information security.Traditional identification technology methods include password verification,smart cards,etc.These technologies are convenient and fast,but they are not very safe,easy to steal,and easy to counterfeit,and they can not meet the current needs well.Human beings have many biological characteristics,such as DNA,fingerprints,voice,iris,human face,and so on.These biological characteristics are unique,reliable,and cannot be simply copied.They can be used as an important basis for identification and verification.Compared with other more mature human biometric recognition methods(such as DNA,fingerprint detection,etc.),face images are easier to obtain,especially in the case of a non-contact environment and without alarming the detected person,the superiority of face recognition far more than other identification technologies.This paper mainly studies image recognition algorithms.Some well-known face databases are used as test objects.Based on the sparse representation principle,two improved face recognition algorithms,IPCR(Improved Probabilistic Collaborative Representation)and KPCR(Kernel Probabilistic Collaborative Representation)are presented.In the AR database test,IPCR and KPCR obtained 98.4% and 98.5% recognition rates,respectively.1.The sparse representation method has the characteristics of strong robustness and high recognition rate,and has been widely used in the field of image classification and recognition.Based on the principle of sparse representation,this paper proposes an improved probabilistic cooperative representation.The sparse coefficient of Po CRC is weighted according to the distance between the test image and each training image.A two-phase method is further used to appropriately select the reconstructed image for recognition training sample set,so it can be classified more accurately.2.In real life,the collected images are often affected by illumination and occlusion,which reduces the recognition rate of the algorithm.Based on IPCR,the KPCR algorithm proposed in this paper,introduces a Gaussian kernel function and replaces low-dimensional linear data with high-dimensional nonlinear data to improve the class interval.A large number of experiments show that the method achieves higher recognition accuracy.