Improved automatic age estimation algorithm using a hybrid feature selection

Abstract

Age estimation (AE) is one of the significant biometric behaviours for emphasising the identity authentication. In facial image, automatic-AE is an actively researched topic, which is also an important but challenging study in the field of face recognition. This paper explores several algorithms utilised to improve AE and the combination of features and classifiers are associated. Initially, the facial image databases are trained and then the features are extracted by employing several algorithms like histogram of oriented gradients (HOG), binary robust invariant scalable keypoints (BRISK), and local binary pattern (LBP). Here, the AE is done in three various age groups from 20 to 30, 31 to 50 and above 50. The age groups are classified by utilising Naïve Bayes classifier (NBC). AE model is calculated by employing the Indian face age database (IFAD) and labelled Wikipedia face (LWF) aging databases obtaining optimistic result with success rate.