A Study on Staircase Detection for Visually Impaired Person by Machine Learning using RGB-D Images

Citations Over TimeTop 15% of 2017 papers

Abstract

There are some previous researches and developments which have purposes to support for visually impaired person. In facts, a survey report about accidents of visually impaired pedestrian said that 42% of respondents experienced walking accidents. Our previous studies had taken an approach of using some image processing techniques to detect the steps of staircases on depth images taken by RGB-D camera. To improve the detection accuracy, this paper applied machine learning techniques, e.g. HOG (Histogram of Oriented Gradients) feature descriptor and AdaBoost classifier, to the staircase detection. This paper shows experimental results of the staircase detection applied to depth-images and RGB-images, and mentions about the effectiveness. Moreover, the classifier AdaBoost and SVM (Support Vector Machine) were compared by a viewpoint about detection accuracy.

Related Papers

• → Extracting depth and matte using a color-filtered aperture(2008)102 cited
• → Enhanced RGB-D Mapping Method for Detailed 3D Indoor and Outdoor Modeling(2016)33 cited
• → Detection of similar objects and localizing on each using Depth Camera(2020)5 cited
• → Partially occluded object reconstruction using multiple Kinect sensors(2014)6 cited
• → Depth Map Super-Resolution for Cost-Effective RGB-D Camera(2015)2 cited