Comparative Study on Face Recognition Based on SVM of One-against-One and One-against-Rest Methods

A comparative study on face recognition based on two methods, SVM of One-against-One method and SVM of One-against-Rest method, has been carried out in this paper. Our method consists of three parts: Firstly, the image robustness on illumination and posture has been improved by the processing of the Gabor wavelet. Secondly, the bilateral 2DLDA technique on image is adopted to realize the dimension reduction of the large feature data. Finally, the SVM classifier has been used to classify facial features and accomplish face recognition. The face recognition is the multiple-classification issue, but the SVM is the essence of handling the binary classification problem. We used the traditional methods of One-against-One SVM and One-against-Rest SVM iteratively to train and test facial database samples to achieve the multiple-classification problems. In the experiment, the ORL face database was applied to compare the two classical methods. The result shows that the One-against-Rest SVM method has the priority of better recognition rate.