Fast and accurate face recognition using SVM and DCT

Abstract

The main problem of face recognition is large variability of the recorded images due to pose, illumination conditions, facial expressions, use of cosmetics, different hairstyle, presence of glasses, beard, etc., especially the case of twins’ faces. Images of the same individual taken at different times, different places, different postures, different lighting, may sometimes exhibit more variability due to the aforementioned factors, than images of different individuals due to gender, age, and individual variations. So a robust recognition system is implemented to recognize an individual even from a large amount of databases within a few minutes. So in order to handle this problem we have used SVM for face recognition. Using this technique an accurate face recognition system is developed and tested and the performance found is efficient. The procedure is tested on ORL face database. Results have proved that SVM approach not only gives higher classification accuracy but also proved to be efficient in dealing with the large dataset as well as efficient in recognition time. Results have proved that not only the training performance, the recognition performance but also the recognition rate raises to 100% using SVM.