Real-time beard detection by combining image decolorization and texture detection with applications to facial gender recognition

There are still many challenging problems in facial gender recognition which is mainly due to the complex variances of face appearance. Although there has been tremendous research effort to develop robust gender recognition over the past decade, none has explicitly exploited the domain knowledge about the appearance difference between male and female. Beard/mustache contributes substantially to the facial appearance difference between male and female and could be a good feature to be incorporated into facial gender recognition. Little work on beard segmentation has been reported in the literature. In this paper, a novel real-time beard/mustache detection method is proposed which combines face feature extraction, image decolorization and texture detection. Image decolorization, which converts a color image to grayscale, aims to enhance the color contrast while preserving the grayscale. On the other hand, beard appearance is normally grayscale surrounded by the skin color face tissue. Hence, it is a fast and efficient way to segment the beard by using the decolorization technology. In order to make the algorithm robust to the variances of illumination and head pose, an adaptive decolonization segmentation has been proposed in which both the segmentation threshold selection and the beard region following are guided by some special regions defined by their geometric relationship with the salient facial feature. Furthermore, a texture-based beard classifier is developed to compensate the decolonization-based segmentation which could detect the darker skin or shadow around the mouth caused by the small lines or skin thicker from where he/she smiles as beard. Only the face is verified as the face contains beard/mustache when it satisfies: 1) a larger beard region can be found by applying the decolonization segmentation; 2) the segmented beard region is detected as beard by the texture beard detector. The experimental results on color FERET database have shown that the pro- osed approach can achieve 89% bearded face detection rate with 0.1% false acceptance rate.