Automatic age estimation based on local feature of face image and regresion

Abstract

With the extensive study on various facial variations, such as illumination, expression gender and age, more and more researchers have realized the importance of the aging effect to a robust face recognition system and engaged in the study on automatic human age estimation based on analysis of human face image. But up to now, most of them focus on the methods based on features extracted from the whole face image. In this paper we develop a framework to estimate human age using the local features from some regions of human face images which are sensitive to aging process. The adaboost algorithm is applied to find the regions we need and LBP features are extracted to model a robust quadratic regression function for age estimation.