Multi-information Fusion Algorithm for Human Target Tracking
Dengtai Tan, Shichao Li, Guangli Wu, Li Denglou, Liping Liu · 2019
A human object tracking algorithm fused with depth information, color information and forecast information is presented in this paper, which enables follow-robot to track motion human object robustly. Firstly, Camshift algorithm is expended into three-dimensional space by color information and depth information. Secondly, the interference from complex background is eliminated from the projection drawing by depth image of motion human object in three-dimensional space. Finally, human object's location information is forecasted by Kalman algorithm and human's center of mass location is calculated in the corresponding depth image, human object's tracking is achieved now. Using Kinect sensor to capture RGB image and depth image, and human object tracking algorithm fused with Multi-information is achieved and applied to the follow-robot. The experimental result shows that the algorithm proposed in this paper can track motion human robustly in condition of complex background, uneven ambient light, deformed target and so on.