Pedestrian crosswalk detection using a column and row structure analysis in assistance systems for the visually impaired

Abstract

Computer vision-based approaches have become more common in assistance systems for the blind and visually impaired where portable devices can be used to assist users in their free movement. The method for pedestrian crosswalk detection with the main goal to facilitate the crossing of the road is proposed in this paper. The proposed crosswalk detection method is based on analyzing the image column and row structure. The Performance of this kind of approach relies on the input image resolution and quality. Therefore, guidance for selecting the appropriate input image resolution for this kind of approach is given. This approach is tested on the realistic input data captured with a monocular camera and using portable devices for image processing.