Research On Gender And Ethnicity Face Recognition Method

Face recognition is one of the most challenging problems in the fields of pattern recognition and computer vision. Recently, it becomes an active topic. As the important consist of face recognition, the face gender and ethnicity recognition is catching an abroad attention and have a broad application foreground.Generally, one gender and ethnicity recognition system consists of three modules, face image preprocessing, facial feature extraction and classifier. This paper researches on the three modules and compares some different methods on gender recognition. Based on researching the theories in face recognition, a new gender recognition method which combined Gabor wavelets, Adaboost learning algorithm and SVM is proposed. About the ethnicity recognition problem, a new combined Gabor wavelets and skin color feature extraction method is proposed.Gabor wavelets transform is similar with human vision characteristic, and is widely used in computer vision and pattern recognition area. In this paper, Gabor wavelets transform is used to extract the facial gender features and also combine it with skin-color information to extract the facial ethnicity features.Adaboost learning algorithm considers the linear classify capability of every feature in each feature vector completely, through the weaklearning process the key features are extracted. Adaboost learning algorithm is used after Gabor wavelet filtering to reduce the dimension of Gabor features and improve the performance of recognition.Support vector machine is based on the theory of structural risk minimization, it has many advantages when to solve the small sample problems, it is already the best classifier in pattern recognition area. In this paper, the Lib-SVM based RBF kernel function is used.In this paper, all the experiments are implemented based on FERET database. PCA+SVM, ICA+SVM, Haar-like+Adaboost+SVM and Gabor+Adaboost+SVM are implemented to recognize the face gender. The experiment results report that the method proposed in this paper gets the better performance and the accuracy is above 90%. In face ethnicity recognition experiments the tree structure SVM is used to classify the Gabor and skin-color combined features and also obtain a good experimental performance.