Facial age classification based on weighted decision fusion

An age classification algorithm based on multi-feature weighted decision fusion is proposed. On the basis of calibration for the facial positive posture, the facial texture features are extracted by three methods. Firstly, the uniform local binary pattern (ULBP) histogram is extracted through the Gabor wavelet transform which has the multi-scale and multi-directional characteristics. Secondly, an ASM-based approach is used to complete facial partition, and the ULBP histogram is extracted from the labeled focus region. Thirdly, the rate between facial wrinkles and facial skin areas is extracted. A strong SVM classifier based on multi-feature weighted decision fusion is designed according to the three features extracted above. The experiments are simulated in the FG-NET and self-build face database for four age-group classification. The results demonstrate the effectiveness of the proposed algorithm. At last, we analyze the impact of the SVM parameters and the face image resolution on the results.