Robust Face Recognition Algorithm Based On Extended Dictionary

Face recognition refers to classification by sampling the feature information on the face and the face image in the database to achieve identity verification.Compared with traditional identity verification,face recognition has more advantages in terms of security,convenience,and cost saving,and improves the quality of life of the people.However,in practical applications,face recognition is susceptible to interference from noise and occlusion under the condition of insufficient sampling,resulting in a low recognition rate.This paper uses the robustness of sparse representation in face recognition,and mainly focuses on the face research under the conditions of insufficient sampling.The specific research contents are as follows:(1)Aiming at the problems of the limited training sample dictionary learning of the traditional face recognition algorithm is unstable,the noise processing is not robust enough,and the running speed is slow,a robust face recognition algorithm with extended dictionary learning is proposed.First use the original training samples to generate two extended training samples;then add random noise with pixel damage as the main influencing factor and structural noise with occlusion as the main influencing factor to the two extended training samples.By increasing the diversity of training samples,More robust dictionary.The experimental results show that the algorithm has high recognition rate,strong robustness to noise,and fast running speed on Extended Yale B,AR,ORL,and Yale databases.(2)Aiming at the problem of random errors in the face recognition process under the condition of insufficient sampling,this paper proposes a weight-based adaptive extended robust dictionary learning algorithm.Firstly,mirror training samples are added to improve the completeness of the dictionary,and then the weight function is used to suppress the influence of outliers on the classification process,and finally the kernel norm is used to coordinate the contradiction between sparsity and relevance.A large number of experimental tests on multiple standard databases of Extended Yale B,AR,ORL and CMU PIE show that the algorithm has a high recognition rate and robustness.