Fusion of Structural and Textural Facial Features for Generating Efficient Age Classifiers

Facial aging is a usual happening that is certain, and varies from person to person depending upon the circumstances and living habits. Applications of age determination are observed in domains like forensic science, security and also to determine health. Facial parameters used for age classification can be either structural or textural. In this paper we have used both approaches for feature extraction. In structural, facial growth is considered for classification, by computing the Euclidean distance between the various landmarks on facial image. The global features used to distinguish child from middle aged and adults is based on the ratios computed using the eyes, nose, mouth, chin, virtual-top of the head and the sides of the face as those features. In textural approach we consider skin texture for classifying the age groups. The prominent areas in facial skin are extracted where significant changes occur in terms of wrinkles that happen in the process of aging. Local Binary Pattern (LBP) feature is used for classification of age in different groups. The experimental results are significant and remarkable.