Multiple face detection algorithm using colour skin modelling

A variety of skin tones along with other constraints on image capturing within a given background make face detection a challenging task of image processing for surveillance and security applications. In this study, a colour-based skin model is proposed for a diverse range of skin tones. It is found that the proposed skin model outperforms several existing skin models. The proposed skin model is used to develop a soft decision-based multiple face detection framework. The robustness of the proposed framework has been evaluated for a wide range of poses, facial expressions, complex backgrounds and blur noises. On the basis of the proposed skin model and the soft decision-based scheme, the range of values of Cb and Cr has been compressed to obtain reduced false detections and missed detections as compared with the existing algorithms.