Scene flow estimation from stereo video source

Abstract

There are various methods to estimate the scene flow. Most of the methods use motion estimation with stereo re-construction. This paper describes an interesting way to fuse the video from two camera's and create a 3D reconstruction. The proposed algorithm incorporates probabilistic distributions for optical flow and disparity. Multiple such re-created renderings can be put together to create re-timed movies of the event, with the resulting visual experience richer than that of a regular video clip, or switching between images from multiple cameras, do a head tracking of the viewer and change the view angle accordingly or view it on a mobile device using the accelerometer for camera tilting for the 3D effect.