Left-right head movement for controlling smart wheelchair by using centroid coordinates distance

Head movement is one stage to handle the movement of smart wheelchair for hands-free disabled person. This tool is useful to help disability of having free movement using computer vision. In this paper, we propose a new framework to recognize left-right head movement by using diagonal coordinates distance (DCD) and horizontal coordinates distance (HCD). Face detection as main part is applied at the first stage and skin detection is combined in terms of failure of face detection. Skin detection stages has some steps as well. First step is image conversion from RGB to HSV and YCbCr channel to get the skin detection to work over face image. From the detected face area, eyes recognition is utilized by using only a half of upper side face’s image. Nose detection is applied as well based on whole face area. The center coordinates of eyes and nose are used to determine the movement referred as centroid. The result shows that HCD is better than DCD. HCD could cope the accuracy 82.5% for head movement detection. This accuracy represents that by using HCD left-right head movement outcome the problem of person with and without glasses as well as the various position of web camera which is mounted in front of smart wheelchair.