Motion Detection and Tracking Using the 3D-camera

Abstract

AbstractThis paper presents how to detect and track the real-time motion using the PMD 3D-camera. Different modalities, including intensity image, range image and amplitude image, can be provided by the 3D-camera. Both the intensity and the range images are used in the motion detection and tracking process. First, intensity and range backgrounds are separately modeled. During the motion detection and tracking, the segmentation results based on intensity and range images are combined, which results in a robust segmentation of the current scene. And such a segmented image has to be filtered in real-time applications because the NIR light emitted by the 3D-camera will be affected by the objects too close to the 3D-camera, resulting in intensity distortion, and low resolution of the intensity image constitutes another reason. After segmentation, edge and contour detection is used to try to figure out the number and positions of mobile objects. By experiments, the process of real-time motion detection and tracking is demonstrated. Finally, some existing problems and proposed solutions such as shadow removal are discussed.