Estimation of longitudinal profile of road surface from stereo disparity using Dijkstra algorithm

• Published in: International journal of control, automation, and systems Vol. 12; no. 4; pp. 895 - 903
• Main Authors: Lee, Ki-Yong, Park, Jeong-Min, Lee, Joon-Woong
• Format: Journal Article
• Language: English
• Published: Heidelberg Institute of Control, Robotics and Systems and The Korean Institute of Electrical Engineers 01.08.2014; Springer Nature B.V; 제어·로봇·시스템학회
• Subjects: Algorithms; Cameras; Control; Control systems; Engineering; Estimates; Estimating techniques; Matching; Mechatronics; Representations; Roads; Robotics; Robotics and Automation; Scene analysis; Studies; Vision systems; 제어계측공학; longitudinal profile of road surface; v-disparity representation; stereo vision; Dijkstra algorithm
• ISSN: 1598-6446; 2005-4092
• DOI: 10.1007/s12555-013-0478-x

A novel algorithm capable of estimating the profile of the road surface along the cameraviewing direction is proposed. When a disparity map from stereo matching is given, the algorithm estimates the profile precisely by applying the Dijkstra algorithm to a v-disparity representation. Although the road surface is very changeable, many algorithms of road scene analysis have relied on flatroad assumption or have estimated the profile roughly. The combination of the Dijkstra algorithm and a v-disparity representation contributes to eliminating such assumption and limit. We demonstrated the validity of the proposed algorithm by performing experiments with many road images.