Improved Face Recognition Rate Using HOG Features and SVM Classifier
Harihara Santosh Dadi, Gopala Krishna Mohan Pillutla · IOSR Journal of Electronics and Communication Engineering · 2016
A novel face recognition algorithm is presented in this paper.Histogram of Oriented Gradient features are extracted both for the test image and also for the training images and given to the Support Vector Machine classifier.The detailed steps of HOG feature extraction and the classification using SVM is presented.The algorithm is compared with the Eigen feature based face recognition algorithm.The proposed algorithm and PCA are verified using 8 different datasets.Results show that in all the face datasets the proposed algorithm shows higher face recognition rate when compared with the traditional Eigen feature based face recognition algorithm.There is an improvement of 8.75% face recognition rate when compared with PCA based face recognition algorithm.The experiment is conducted on ORL database with 2 face images for testing and 8 face images for training for each person.Three performance curves namely CMC, EPC and ROC are considered.The curves show that the proposed algorithm outperforms when compared with PCA algorithm.