A Hybrid Face Detection Approach Based on Facial Features and Skin color Information

Abstract

Face is the primary identity of an individual and the differences in faces of individuals can be used in automatic visual identification system for identification of people. This work is based on an improvement of Viola-Jones face detection in color images. A hybrid face detection approach is employed by merging the Viola-Jones face detection technique with the skin detection technique. It has been observed that the performance of Viola-Jones detection technique is mostly affected due to background complexity, non-standard illuminations, camera quality, distance from camera, multiple faces in the image, and variations in poses and expressions and because the facial features are not accurately extracted in such conditions. In such cases, facial skin information is provided as an additional feature to Viola-Jones classifier, on the base of which face detection is carried out. The experiment showed that the proposed hybrid approach outperforms the Viola-Jones face detection technique.