Head posture detection using skin and hair information

This paper addresses a novel head posture detection algorithm to recognize human-computer interactions. A pattern training based image segmentation algorithm is used to detect the skin and hair of students. A simple and efficient human presence detection and gaze direction estimation method is then proposed based on the segmentation results. Finally, the proposed algorithm is tested on ten different students for seven different head postures each. The experimental results show that 92% of head posture images are accurately identified.