Fatemeh Ghorbani Kotak - Islamic Azad University of Buin Zahra, Faculty of Engineering, Department of Computer Engineering, Buin Zahra, Iran
Abbas Koochari - Islamic Azad University of Buin Zahra, Faculty of Engineering, Department of Computer Engineering, Buin Zahra, Iran

Aging is the process of appearing of major variations on human face such as changes in shape, texture, facial hair and etc. Because of occurring of these variations in specific ranges of age, human age can be estimated automatically according to these variations. A near research studies to automatic age estimation is age simulation where the facial image of individual is simulated for a selected test age value. One of the main challenges of age simulation approaches is feature extraction in which almost all facial feature extractors depend on face geometry and so, changing the facial features over the aging process affects reduces the accuracy of image estimation system. To solve the problem, this paper proposes the utilization of Pseudo Zernike Moment (PZM) as a novel feature extraction method in age simulation system. Since PZM is robust to the shifting, scaling and transformation of object inside the image, it is able to efficiently extract feature vector even in the case of facial variations over the aging process. Also, the proposed method uses the Multi-Layer Perceptron (MLP) as the learning method to simulate the face image of individual in a specific age value. Evaluation results of the proposed method on FG-NET and MORPH datasets proves the superiority of the proposed method to the other similar methods

Age simulation, Pseudo Zernike Moment (PZM), Feature extraction, Multi-Layer Perceptron (MLP)