Sex Based On Facial Image Classification

Face Image contains a wealth of characters, including gender, age, race, identity and so on. Gender classifying from facial image is to find the face from image and determine whether it is a male or female. A gender recognition often comprises three modules:face detection, face feature extraction and gender classification. In this paper, we did research on these three parts.In face detection, we researched the training and detection of AdaBoost algorithm. Considering the image quality, we took image preprocessing firstly. After detection, we found some misdetection in the results. So we used a method to remove them. It is that we continue to find the locations of eye, nose and mouth in the image. And we determine whether the result contains face with the relative positional relationship of these parts.In feature extraction, we studied the PCA algorithm and LDA algorithm. And, we compared them in image processing with two experiments. Finally, we combined them to recognize the gender.Finally, we gave a thoroughly introduction on the basic concepts on ANN and CNN fields, including the algorithm we used in work. We adapted the CNN to gender classification. We constructed a basic CNN model, and applied them to different face dataset. And we constructed some models with different number of feature extractors based on the basic model. By comparing the experimental result of Misclassification Rate, we analyzed the models that have different structure.