A Robust Feature Extraction Method with Pseudo Zernike Moment in Conditional Probability Neural Network-Based Automatic Facial Age Estimation

Human face is one of the most important features of body which varies significantly over the aging process. Due to the reason, the task of facial age estimation can be performed on human face. In the task of age estimation, feature extraction is the first important step which highly effects on training of pattern recognition method and obtained results. The second important step of an age estimation system is training of pattern recognition method based on the extracted feature vector. According to the importance of feature extraction and training steps, this paper proposes the combination of Pseudo Zernike Moment (PZM) and Conditional Probability Neural Network (CPNN) as an efficient and robust facial age estimation approach. In this paper, PZM as the feature extraction method is employed to extract full informative feature vector elements even in the case of deformation, scaling, rotation and transformation of face in image. The utilized CPNN acts as a learning method to predict the age value of a particular face image based on the extracted feature vector. The results of the propose method on FG-NET dataset proves the superiority of the proposed age estimation method to the other age estimation methods which are failed to predict the age value in almost all forms of face inside the image such as scaling, shifting, rotation and transformation.