Motion estimation for stereoscopic video and its application to frame rate up-conversion for 3-D TV

To display video contents vividly on consumer devices including television (TV) and computer, frame rate up-conversion (FRUC) has been widely employed. However, for stereoscopic video, applying any conventional FRUC algorithm to each view may cause severe annoying artifacts resulting from inconsistency between two motion vector fields. In this paper, a novel motion estimation (ME) algorithm is proposed for stereoscopic video and applied to FRUC for three-dimensional TV (3-DTV). By searching motion for correspondences in stereo views simultaneously, reliable and consistent motions having only horizontal disparity can be obtained, and much higher visual quality and depth representation are delivered when the interpolated stereoscopic video is viewed.