Embedded Identification System Based On Face Recognition
Issues with thefts and identity fraud are countinously increasing and becoming more and more serious as the usage of computerized systems for identification purposes is increasing. In this aspect, for controling access in private buildings, it becomes a necessity to use more robust techniques than those that are commonly used nowadays. Face recognition is one of the techniques that can increase the security. In this paper, we present an embedded implementation of a security access control system based on face recognition. Haar-like features are used for feature extraction, the PCA algorithm is used for eigenvector calculation and SVM based classification is used for face recognition. Our goal was to evaluate the feasibility and efficiency of implementing such a system on low-price embedded devices. This paper focuses on evaluation of the systemís accuracy and how controlling parameters such as background, light and number of trainings influences this accuracy.
Index Terms- Face Recognition, Feature Extraction, Classification, Embedded Devices, PCA, SVM.