Research On Face Recognition Based On DCT And SVM

With the urgent need for identity verification in daily life, the researchers are increasingly concerning about the Face Recognition. Face recognition has great potential prospects. It not only can be used after the track, but also prevention and control and many other real-time applications.In order to solve face images of the high dimensionality, we use discrete cosine transform to extract the main features from the whole face images. Compared to other methods, it has great advantage for speed. While only a small number of DCT coefficients can be effectively characterized by sample data. Thereby it improves the speed of the whole recognition process.In this paper, we use support vector machine for classification of face recognition. Because face recognition is a typical problem of small sample size, and SVM is to solve exactly the problem of small sample out of the way. This article study and analysis deeply of the SVM method and discusses the advantages and disadvantages of multi-classification, and discusses in detail the aggregation Fuzzy algorithm and Least Squares Support Vector Machine theory. The purpose of introducing them to further improve the speed, and ultimately achieve the speed requirement of real-time applications. We done a lot of experiments in ORL face database and showed that this method indeed improved speed.