Smartphone-based crosswalk detection and localization for visually impaired pedestrians

摘要: This paper describes recent work on the “Crosswatch” project, which is a computer vision-based smartphone system developed for providing guidance to blind and visually impaired travelers at traffic intersections. A key function of Crosswatch self-localization - estimation user's location relative crosswalks in current intersection. Such information may be vital users with low or no vision ensure that they know crosswalk are about enter, properly aligned positioned crosswalk. However, while methods have been used finding helping align themselves them, these assume entire pattern can imaged single frame video, poses significant challenge user who lacks enough where point camera so as In this we describe progress tackles problem detection self-localization, building describing techniques enabling acquire 360° image panoramas turning place sidewalk. The panorama converted an aerial (overhead) view nearby intersection, centered standing at, facilitate matching template intersection obtained from Google Maps satellite imagery. process allows features detected permits precise interest. We demonstrate our approach imagery acquired by users, thereby establishing feasibility approach.