Facial Features Extraction Using LBP for Human Age Estimation Based on SVM Classifier

Abstract

Research on age estimation witnessed increasing attention due to the demand for its applications. The age estimation has an essential role in preventing under-age persons from performing adult activities. The proposed age estimation technique is carried out through several stages; preprocessing, feature extraction and then age classification. In this paper, the Local Binary Pattern (LBP) algorithm is adopted to extract the face features focusing on selecting the best possible combination among all the features produced from the LBP algorithm. Feature Selection Method (FSM) is employed to increase the accuracy. FSM yields better results compared to other techniques’ results. Support Vector Machine (SVM) is used to classify the tested person image and assign that person to the related age. Results conducted using MATLAB produced accuracy of 93.81% with FSM technique compared to 81.61% without it. When damaged images are excluded from the database used for training, the accuracy is increased to 94.57%.