algorithm - 3d model construction using multiple images from multiple points (kinect) -
is possible construct 3d model of still object if various images along depth data gathered various angles, thinking have sort of circular conveyor belt kinect placed , conveyor belt while real object reconstructed in 3d space sits in middle. conveyor belt thereafter rotates around image in circle , lots of images captured (perhaps 10 image per second) allow kinect catch image every angle including depth data, theoretically possible. model have recreated textures.
what know whether there similar projects/software available , links appreciated whether possible within perhaps 6 months how proceed this? such similar algorithm point me , such
thanks, milindad
it possible , there lot of 3d scanners work out there, more or less same principle of stereoscopy.
you know this, contextualize: idea 2 images same point , use triangulation compute 3d coordinates of point in scene. although quite easy, big issue find correspondence between points in 2 images, , need software extract , recognize similar points.
there open-source project called meshlab 3d vision, includes 3d reconstruction* algorithms. don't know details of algorithms, software entrance point if want play 3d.
i used know other ones, try find them , add them here:
(*wiki page has no content, redirects login editing)
Comments
Post a Comment