Tour Into The Picture

The goal of this project is to create a simple, planar 3D scene from a single photograph. The problem will follow the description in Tour into the Picture by Horry et al. in modeling the scene as a 3D axis-parallel box. First, we will let the user specify simple constraints on that box (the back wall plus the vanishing point). Then, its just a matter of extracting the coordinates of the box in 3D, and texture-mapping the faces of the box. The paper has a rather poor description of the process; implementation was actually pretty straightforward. Here's the paper: Youichi Horry, Ken-Ichi Anjyo, and Kiyoshi Arai. Tour into the picture: using a spidery mesh interface to make animation from a single image. In Proceedings of the 24th annual conference on Computer graphics and interactive techniques, pages 225{232. ACM Press/Addison-Wesley Publishing Co., 1997.

Notes:

- 3D geometry estimation of the room: The 3D coordinates of each vertex of each of the five planes are used to define the 3D geometry corresponding to these planes.

- For each of the five planes, we find the transfomation that will take the camera from the view that it sees in the image to a fronto-parallel view of the plane (which is a view that looks at the plane head on). In this case, the transformation will just be a homography, H, were H is a 3 by 3 matrix which relates the source and destination points, s and d (in homogenous coordinates), as d = Hs. This is simply a system of equations using the correspondence between the source and the destination points to obtain H.

- Apply homography to generate fronto-parallel views: visualizing the fronto-parallel view of the floor plane gives us tile lines that appear parallel.

- After conversion to a 3D model of the room, new viewpoints of the scene of obtainable.

- Focal length is estimated geometrically.

All images are produced from the program. As you can see, after such transformations, we can create new viewpoints of the scene from only one image.