Research And Application Of Face Recognition Via Sparse Representation

Face is one of the most significant features of human body, and it is the basic approach to distinguish one person from another. As a challenging problem related to many subjects, face recognition algorithms are booming in recent several decades. However there are still many difficult problems that need to be solved. With the development of sparse representation and compressed sensing theory in the last decade, there emerged a new face recognition algorithm called sparse representation-based (SR) classification, and researchers conducted some experiments on the standard databases to verify the efficacy of this algorithm. Then some other researches made further study and improvement on the theory, and proposed new algorithms like collaborative representation-based (CR) classification and extended sparse representation-based (ESR) classification, and they both achieved higher recognition rate on the standard databases.This paper proposed a new algorithm called extended collaborative representation-based (ECR) classification based on the above-mentioned algorithms, it makes up the undersampled problem of collaborative algorithm and the computation complexity of the extended sparse representation algorithm, and then I made a comparison of ECR algorithm with CR algorithm and ESR algorithm under same experimental conditions, and achieved the highest recognition rate.In the last part of this paper, I tested and discussed the practicability of ECR algorithm using daily photographs, at the same time I tried to find out the most practical algorithm by comparing the recognition rate and running time of these different algorithms.