Face gender recognition is to discriminate the gender based on the inputted face image, which is a typical two-class pattern recognition problem. In this paper, the research work about the face gender recognition can be summarized as follows:(1) With Support Vector Machine classifier, four general feature extraction methods are compared in experiments, included Principal Component Analysis, Local Binary Pattern operator, multi-scale Local Binary Pattern operator and Gabor wavelet feature;(2) With Gabor wavelet feature and Support Vector Machine classifier, three facial image masks are compared in experiments;(3) A face gender recognition method based on cascade Support Vector Machine is proposed, which can remove easily classified samples by pre-layer classifier, and train the new layer classifier by re-organizing the difficult training samples. This idea of multi-layer classifier can effectively improve the recognition performance;(4) Classifier fusion method applied on face gender recognition is researched, which can enhance the robustness and generalization capability;(5) Modification, optimization and parallelization on LIBSVM algorithm is researched, which can reduce the training time significantly;(6) Design and implementation of the face gender recognition system. |