Assistance Instruments Detection Using Geometry Constrained Knowledge for Health Care Centers

A walking frame and wheelchair detection system that uses geometry constrained knowledge is proposed for health care centers. The system can early detect an accident occur when the elderly get on bed. Based on the space relative constrained, accurate results are obtained for wheelchair position detection. A tracking algorithm is utilized to track and predict wheelchair motion with orientation. The developed approach has been applied for the early detection accidents in a health care center. Experiment results show that the algorithm obtains desirable performance and accuracy.