A method for acquiring 3D shapes for silhouettes generated by multiple color light sources

摘要

Volume intersection is known as a method for acquiring a 3D shape of an object from multi-view images. When the object is moving, the method generally requires plural cameras to obtain synchronized multi-view images. This paper proposes a kind of volume intersection methods which employs only one camera and multiple color light sources. At first, we estimate light source disposition and camera parameters by measuring shadows of known objects projected onto a floor. Next, a shadow image of an unknown object is taken by the camera and shadows which correspond to individual light sources are separated from one another based on their colors in the image. Finally, a 3D shape of the object is extracted as a common part of volumes defined by the light sources and their respective shadows.