Abstract
Markerless human motion tracking can be employed in many applications such as automatic surveillance, motion capture, human-machine interface and activity recognition. This problem has been extensively studied in the computer vision research community in the last years. In this context, the present paper presents an approach for 3D markerless human motion tracking based on a skeletal kinematic model of the human body. This method is applied over a 3D probabilistic occupancy grid of the environment, which is constructed by means of a Bayesian fusion of images from multiple synchronized sensoring cameras. Although the use of kinematic models in 3D human tracking is widely employed, its use over 3D probabilistic occupancy grids still was not vastly investigated in the literature. The experiments were performed using a public dataset with video sequences of people in motion. The results show that the method is capable of dealing adequately with the 3D markerless human motion tracking problem.
Highlights
Visual tracking, which is the recursive detection and location of objects in videos [1], is a classical computer vision problem
One could mention automatic surveillance, motion capture, human-machine interface and activity recognition, as examples of relevant problems where the visual tracking of people is employed. Tracking articulated targets, such as the human body, using 2D images is a difficult problem to be treated mainly due to: i) the complex nature of 3D movements; ii) the loss of information in images because of 2D space restriction; iii) the color changes caused by luminosity variations; iv) the existence of others objects moving into the scene
The results show that the method is capable of dealing adequately with the 3D markerless human motion tracking problem
Summary
Visual tracking, which is the recursive detection and location of objects (or more generally, visual patterns) in videos [1], is a classical computer vision problem. One could mention automatic surveillance, motion capture, human-machine interface and activity recognition, as examples of relevant problems where the visual tracking of people is employed. Tracking articulated targets, such as the human body, using 2D images is a difficult problem to be treated mainly due to: i) the complex nature of 3D movements; ii) the loss of information in images because of 2D space restriction; iii) the color changes caused by luminosity variations; iv) the existence of others objects moving into the scene. From the set of images captured by the cameras a 3D reconstruction of the environment can be performed
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
