VRML animation from multi-view images

In this paper, we propose a VRML animation system for real objects. This system creates a VRML model from images captured by multiple cameras surrounding the target object. In order to create 3D models of humans, the correspondence detection method is used with the volume intersection method as a restriction condition in the matching process. By using this method, a relatively accurate human model is generated. We show the result of generating a VRML animation by consecutively displaying VRML models. It is confirmed that the VRML animation can be smoothly displayed on the VRML browser.