We have just started using skanect but we have some problems. Our goal is to scan a room and align the 3D model with the tracking of a VR headset. To do this we use SVD but this gives bad results because the room model, which is passed to the SVD, acquired with the kinect has some deformation, in fact the ratio between the distances in the 2 reference systems is not constant. Our referent says that probably the kinect produces deformation which increase with the distance but we don't know if it's true.
Does anyone know a good way to avoid distortion during the acquisition of a room? Thanks for attention,
PS: we are using a bounding box's size of 5x5x5 and we start scanning from the center of the room.