3D Image Reconstruction of a Cube from Multiple 2D Views
This Demonstration reconstructs 3D images of cubes given multiple 2D images with a chosen mode, either manual or automatic. In manual mode (snapshots 1 and 2), you select points (e.g. corners of the cubes) manually by dragging the locators. The algorithm then reconstructs the image from the selected points.[more]
In automatic mode (snapshot 3), the algorithm finds the points itself thanks to the built-in Mathematica image processing function ImageFeatureTrack.[less]
Here is a brief description of the algorithm.
1. The method for the 3D reconstruction uses a vector approach based on a pinhole camera model.
2. The input consists of two or more images, the position of the cameras used to take the pictures relative to each other, and some information about the camera (e.g. focal length, image size).
 R. Hartley and A. Zisserman, Multiple View Geometry in Computer Vision, New York: Cambridge University Press, 2004.