A Method to Track and Acquire the 3D Point Cloud Data of Object

In the paper, a method to track an object and acquire the 3D point cloud data of the object is proposed, because real-time tracking of an object is challenging work. A robust Bayesian framework is designed in this method for visual tracking of human motion. In this paper, a 3D bounding box is exploited to acquire the spatial range of the human body. Besides, the mapping of the coordinate relation between RGB image and point cloud data is realized through the camera calibration. The experiment results demonstrate that the method can offer a faster convergence of estimate. In addition, the prior distributions of the particles at each sample are defined based on the color histograms of the human body.