Wheelchair Detection in a Calibrated Environment
Ashish Myles, Niels da Vitoria Lobo, Mubarak A. Shah · 2002
This paper describes a method of detecting wheelchairs in 3D for automated assistance for the disabled. It pre-processes each frame by using background subtraction to remove irrelevant data. It then uses skin detection and a variation of the Hough Transform to locate the face of the person sitting in the wheelchair and the wheels of the wheelchair, respectively. After the raw data output from the previous steps is pre-processed, this information is combined with the parameters of the calibrated background to calculate the world coordinates of the wheelchair.