Human age classification using facial skin analysis and multi-class support vector machine

Human age classification via face images becoming an interesting research area because of potential applications in the field of computer vision such as Age Specific Human Computer Interaction (ASHCI), biometrics, security and surveillance, etc. In this paper, a novel method for human age classification using facial skin analysis for aging feature extraction and Multi-class Support Vector Machine (M-SVM) for age classification is proposed to classify the face images into four age groups. Facial skin analysis consists of skin texture analysis and wrinkle analysis. Gabor wavelet is used to analyze the facial skin textural changes with age progression. Wrinkle analysis detects the wrinkle density changes at particular regions on face image with age progression. The performance evaluation of proposed age classification system is carried out by using face images from PAL face database. In this paper, the performance of M-SVM classifier is compared with the performance of Artificial Neural Network (ANN) classifier for the task of human age classification using Gabor wavelet and wrinkle analysis. The result analysis concludes that the best age classification accuracy of 93.61% is achieved by using proposed age classification system and M-SVM is an efficient classifier than the ANN classifier for the task of human age classification in combination with Gabor wavelet and wrinkle analysis.