Local binary patterns and wrinkle analysis in combination with multi-class support vector machine for human age classification
Manesh Kokare, Jayant Jagtap · International Journal of Applied Pattern Recognition · 2017
The ageing patterns reflected on the human face with age progression can be used directly to develop age-based intelligent systems for the applications such as biometrics, security and surveillance, etc. This paper presents a novel human age classification system based on local binary patterns (LBP) and wrinkle analysis for ageing feature extraction and multi-class support vector machine (M-SVM) for age classification to classify the face images into four age classes. In this paper, the potential of LBP to capture the minute variations in the textures has been exploited to extract an efficient and robust ageing features invariant to illumination and rotation changes. The proposed age classification system is trained and tested with the face images from PAL face database and achieved the improved age classification accuracy of 91.39%. The experimental results comparison of MSVM classifier with nearest neighbour (k-NN) classifier and artificial neural network (ANN) classifier concludes that M-SVM classifier is the best classifier among the three classifiers for the task of human age classification using LBP and wrinkle analysis for ageing feature extraction.