A real-time smartphone-based floor detection system for the visually impaired

Abstract

According to the American Foundation for the Blind (AFB) more than 25 million people in the U.S. suffer from total or partial vision loss. Assistive technologies have been a pivotal tool to enhance blind people's lives in the last 20 years. Fall prevention (FP) is a research area that has been active for over a decade to improve people's lives through the use of pervasive computing. This work introduces a smartphone-based fall prevention system for the blind and elaborates on the first module: a floor detection system for indoor environments using a smartphone's camera. Image-based floor detection encompasses multiple stages that makes the entire process remarkably difficult. This difficulty is increased due to the complexity of current algorithms, the limited amount of resources available in mobile devices, the movement of the camera while walking, and the real time nature of the system. This paper provides a general description of the fall prevention system along with its challenges and current solutions. Then, a detailed description of the floor detection system is provided including its five modules: smoothing, edge detection, line detection, wall-floor boundary detection, and floor detection. Finally, the floor detection module evaluation shows an accuracy of 82%, a precision of 90.3%, and a recall of 75%.