Merging Images from Parallel Depth Cameras

In this paper a problem of simultaneous information acquisition from multiple depth cameras is investigated, aiming at obtaining single overall picture containing information from all cameras. The experiments are carried out on Microsoft Kinect devices. A methodology for merging images from multiple positioned in a line cameras is proposed. The method is based on the concept of simulating a view of an imaginary camera covering the field of view of depth cameras. For each considered pixel from cameras' views the position in imaginary camera view is established by using geometric properties. As exemplary usage merging of user masks is presented in the paper. The method could be as well utilized for merging locations of other specific points like detected user joints, or even full depth scenes.