A real-time eyebrow segmentation and tracking technique to support an electric wheelchair interface

This paper presents an eyebrow tracking method to support an interface for an electric wheelchair. This interface aims to drive a wheelchair by interpreting the movement of the head and the facial features as well, without hand generated commands. Hardware for the control interface is composed by the helmet with a webcam attached to it: the camera is pointed to the face of the user and the image sequence is acquired and processed in real-time. This paper focuses on the interpretation of commands given by the eyebrows movements. Following the detection of the eyes regions, eyebrows are segmented by a composition of the CIELab colorspace L and b bands, binarized by the classical Otsu thresholding. Tracking is done by computation and analysis of the vertical bit signature, extracted from the segmented eyebrow. The generation of move and stop commands have produced satisfactory results. The eyebrow tracking demonstrated to be accurate and robust in trepidation and different light conditions and users skin tones.