Depth map and color images based on treadmill game interaction system
|School||Huazhong University of Science and Technology|
|Course||Communication and Information System|
|Keywords||Depth map Continuously adaptive mean shift algorithm Human-Computer Interaction Tracking|
Somatosensory game sport though swept the world , but its effect is still far less than the normal exercise treadmill . The game system if the treadmill somatosensory game action and treadmills combine organic , monotonous exercise will give unprecedented entertainment pleasure, so that more users can consciously take the initiative to participate in sports in the past. Because of the non-contact and direct, image recognition technology has become one of somatosensory research focus , in particular, unmarked recognition technology that allows the human body without any additional accessories that become a \After a presentation of the main tracking algorithm, we need to choose according to the system of continuous adaptive mean shift algorithm (CAM-Shift) Tracking hands . Experiments show that the conventional continuous adaptive mean shift algorithm to track the effect is unsatisfactory bare hands in a complex context basically can not work properly. Although the pair of gloves tracking better, but contrary to the system design requirements unmarked . To this end , in the traditional continuous adaptive algorithm based on mean shift , we have introduced depth information, and use the depth map from the foreground segmentation, initialization and interference handling aspects of the algorithm has been improved and expanded. This article is a depth map thresholding to obtain a binary mask of foreground objects , mask out the background in order to reduce the complexity of the search results of non-target areas the impact of interference , while taking advantage of the 3D information depth map initialization search window. On this basis, build depth map and color images based on treadmills interact with the game system . System using Microsoft's Xbox Kinect somatosensory peripheral device for moving the player's depth image and color image sequence, and the trajectory of the players hands for 3D tracking, analyzing identify their hand movements to achieve a non-contact , non- labeled human interaction. Experiments show that the system meet real-time requirements, the action responsive and accurate.