Downstairs Detection for Visually Impaired People using Illuminance of Point Patterns

Abstract

Existing research have proposed the detection of downstairs on routes for visually impaired people. The existing system takes 0.6 seconds to detect the downstairs within 1.3-meters using hough transform method and stereo matching method. However, visually impaired people can't safely go down the stairs using existing systems. Because visually impaired people have to prepare to go down the stairs immediately before placing one's foot on the downstairs. This paper presents high speed detection method which detects the boundary line between the downstairs and the floor using the luminance of point pattern. Experiments results have shown that the proposed method detects the downstairs within 2-meters with an accuracy rate of 90.0%. In addition, detection time is 0.17 seconds.