DocumentCode :
3602253
Title :
Real-Time Multipedestrian Tracking in Traffic Scenes via an RGB-D-Based Layered Graph Model
Author :
Shan Gao ; Zhenjun Han ; Ce Li ; Qixiang Ye ; Jianbin Jiao
Author_Institution :
Sch. of Electron., Electr. & Commun. Eng., Univ. of Chinese Acad. of Sci., Beijing, China
Volume :
16
Issue :
5
fYear :
2015
Firstpage :
2814
Lastpage :
2825
Abstract :
Multipedestrian tracking in traffic scenes is challenging due to cluttered backgrounds and serious occlusions. In this paper, we propose a layered graph model in image (RGB) and depth (D) domains for real-time robust multipedestrian tracking. The motivation is to investigate high-level constraints in RGB-D data association and to improve the optimization from the trajectory level to the layer level. To construct a layered graph, we define constraints in the depth domain so that pedestrian objects in the image domain are assigned to proper layers. We use pedestrian detection responses in the RGB domain as graph nodes, and we integrate 3-D motion, appearance, and depth features as graph edges. An online updating depth factor is defined to describe the depth relationships among the observations in and out of the layers, and the occlusion issue is processed with an analytical layer-level strategy. With a heuristic label switching algorithm, multiple pedestrian objects are optimally associated and tracked. Experiments and comparison on five public data sets show that our proposed approach significantly reduces pedestrian´s ID switch and improves tracking accuracy in the cases of serious occlusions.
Keywords :
feature extraction; graph theory; image colour analysis; image motion analysis; object detection; object tracking; optimisation; pedestrians; 3D motion features; RGB domain; RGB-D data association; RGB-D-based layered graph model; analytical layer-level strategy; appearance features; cluttered backgrounds; depth domain constraints; depth features; depth relationships; graph edges; graph nodes; heuristic label switching algorithm; high-level constraints; image domain; layer level; online updating depth factor; optimization; pedestrian ID switch; pedestrian detection; pedestrian objects; real-time multipedestrian tracking; serious occlusions; tracking accuracy; traffic scenes; trajectory level; Cameras; Data models; Feature extraction; Linear programming; Real-time systems; Target tracking; Trajectory; Multi-pedestrian tracking; RGB-D data; layered graph model; occlusion;
fLanguage :
English
Journal_Title :
Intelligent Transportation Systems, IEEE Transactions on
Publisher :
ieee
ISSN :
1524-9050
Type :
jour
DOI :
10.1109/TITS.2015.2423709
Filename :
7106548
Link To Document :
بازگشت