Skip to Main Content
Human age estimation using facial image is becoming more and more investigated because of its potential applications in many areas such as multimedia communication and human computer interaction. Since many factors contribute to the aging process like gender, race, health, living style, the current age estimation performance for computer vision systems is still not efficient enough for practical use. In this paper, we addressed the problem of age estimation from single facial gray-scale image since the color information appeared as not significant in considered low resolution images. Local and global Discrete Cosinus Transformation (DCT) are used for feature extraction allowing thus a first dimensionnality reduction through this discriminative representation. A second reduction of dimensionality has been obtained through principal component analysis(PCA). A linear regression function has been learned and tested on different large databases extracted from MORPH. Experimental results have shown some encouraging results.