Multiple Feature Extraction and Multiple Classifier Systems in Face Recognition

TL;DR: The results obtained show that this research significantly boosts the accuracy of face recognition in contrast with previous methods.

Abstract: Nowadays, face recognition is one of researchable issues in machine vision. In this research a new method using Multiple Feature extraction and Multiple Classifier system (MFMC) has been presented for face recognition. At First, images are gathered from Cohn–Kanade database and segmented via masks. Then image features are extracted via Local Binary Pattern (LBP), gradient histogram and masks. For classifying extracted features, Naive Bayes, K Nearest Neighbors and Support Vector Machine classifiers are employed by using MFMC Classifier system. The results obtained show that this research significantly boosts the accuracy of face recognition in contrast with previous methods. The proposed method achieved 99.63% recognition accuracy on Cohn–Kanade database.