Scene reconstruction from kinect motion

This paper demonstrates the capabilities of the Kinect [1] device for the purpose of building dense 3D map of the small indoor environments. The range scans from this device provide the information about the 3D structure of scene in the form of 3D point clouds. The alignment problem of these 3D points is solved by tracking the camera movement using the computer vision algorithms, so the exact camera position and rotation known in every time frame can be used to reconstruct a consistent map from multiple Kinect depth images. The purpose of this method is to effectively produce dense 3D maps of small workspaces.