Age Estimation, A Gabor PCA-LDA Approach

Abstract

Automatic human age estimation has considerable potential applications in human computer interaction and multimedia communication. In this paper the Gabor wavelet and its characteristics as a powerful mathematical and biological tool, was used for feature extraction. A combination of Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) was used to reduce dimension and enhance class separability. Finally Euclidean distance was used to classify the images into one of three major groups. These groups are: Group1 (0 to 3 years), Group2 (5 to 10 years) and Group3 (20 to 80 years). The robustness and accuracy of the proposed system was tested on the FG-NET [1] and MORPH [2] public face aging databases. This system was able to achieve 90% accuracy.