Gender Recognition And Age Estimation Based On Facial Features

Human face is an important bio-feature. It’s complicated in structure and detail, yet it contains a quantity of information, such as gender and age. The human face recognition is well developed, while the predictions of face properties, as age and gender, are not yet well studied. This work designs an approach for gender classification and age estimation of human face based on HOG features.The HOG feature, which have succeeded in the application of pedestrian detection, can de-scribe the shape and appearance of an image according to the gradient of each pixel value. Com-pared to other feature descriptors, HOG is robust to the change of the illumination and image geom-etry. More important, HOG feature can well describe the high level and middle level face features, as shape and texture, but ignore the changement of low level features, as the value of the pixel. So HOG can be well applied on the application of human face. HOG feature is also easy to understand and lite on computation.The main work of this article are as follows:1. Inspired by the Human Visual System, we propose an approach based on the fusion of multi-scale HOG feature. The experiments prove the multi-scale HOG feature performs better than simple-scale HOG feature on the application of gender classification.2. We propose a new feature named weighted global HOG by improving the HOG feature. This new feature can greatly decrease the dimension of features by applying PCA. The results of our experiments on MORPH database shows that the new feature performances well on age estimation.3. Based on quantities of experiments, we give some suggested values of parameters of HOG during the application on human face.