Alignment of human body and scenes in 3D reconstruction

In recent years, virtual human technology has been developing fast, and accordingly, the 3D reconstruction of the human body and scenes is required in many applications. Usually, the human and the scene are rendered separately before merging them during the reconstruction, which thus requires a coordinate alignment between the human body model and the scene model. However, this alignment still depends on manual setting in most cases and sometimes significant deviations can occur if the setting is not reasonable. To solve this issue, we propose a new method to adaptively make alignment of the human body and the scene by using a natural physical constraint between these two. We first construct the ground plane model as well as the plane model that the feet of human body stands on. and then make these two models alignment. Experiments illustrate that our method is effective and adaptive, showing the potential to be used in practice.