Automatic Human Age Estimation Using Overlapped Age Groups

Abstract

Facial aging effects can be perceived in two main forms; the first one is the growth related transformations and the second one is the textural variations. Therefore, in order to generate an efficient age classifier, both shape and texture information should be used together. In this work, we present an age estimation system that uses the fusion of geometric features (ratios of distance values between facial landmark points) and textural features (filter responses of the face image pixel values). First the probabilities of a face image belonging to each overlapping age groups are calculated by a group of classifiers. Then an interpolation based technique is used to produce the final estimated age. Many different textural features and geometric features were compared in this study. The results of the experiments show that the fusion with the geometric features increases the performance of the textural features and the highest age estimation rates are obtained using the fusion of Local Gabor Binary Patterns and Geometric features with overlapping age groups. © Springer-Verlag Berlin Heidelberg 2013. © 2017 Elsevier B.V., All rights reserved.