On Creating Depth Maps from Monoscopic Video using Structure from Motion

Abstract

Abstract — The depth-image-based rendering technique is a promising technology for three-dimensional television (3D-TV) systems. For such a system, one of the key components is to generate a high-quality per-pixel depth map, particularly for already existing 2D video sequences. This paper proposes a framework for creating the depth map from uncalibrated video sequences of static scenes using the Structure ¿From Motion (SFM) technique. This paper describes the architecture and the main components of the proposed framework. The initial experimental results show that SFM can be an effective way for creating the depth map, or it can be used to refine the depth map created by other methods, for example, the Depth From Cues (DFC) technique. I.