2D to 3D conversion for close-up soccer Highlights

2D to 3D conversion for close-up soccer Highlights

Speaker:

Anis Troudi

Abstract:

A wide spread adoption of 3D stereoscopic television and the coming wave of virtual reality devices are hindered by the lack of high-quality 3D and immersive content. One promising solution to address this need is to use automated 2D-to-Immersive content conversion.

Current conversion methods, while general, produce low-quality results that exhibit artifacts that are not acceptable to many viewers. We address this problem by showing how to construct a high-quality, domain-specific conversion method for sports videos. Our method is data-driven — we generate stereoscopic frames for a given video stream by transferring depth information from a database of similar 3D stereoscopic videos. First we generate segmented maps (binary images) of players after analyzing the motion of the highlight sequence, Once we retrieve similar 3D video frames, our technique transfers depth gradients to the target frame while respecting object boundaries using the segmented maps, solves for the output depth map, and then generates stereoscopic video, we also applied inpainting technics to improve objects disparity (pop-up) to give more immersive 3D view experience out of 3D TVs and VR devices.