Multiple View Pose Estimation

Computer Graphics and Multimedia Laboratory’s project by Yuval Nirkin and Itay Shvartz, supervised by Michael Kolomenkin.

 

The Project goals:

·       Given a set of pictures taken by a single calibrated camera, find the relative position and rotation of the camera that corresponds with each of the pictures.

·       Create software with a user interface that allows proper selection of the input pictures and pairs by the user.

·       Render a 3D scene of the cameras and images.

 

Motivation:

Being able to retrieve the relative positions and rotations of calibrated cameras is a crucial step in performing the following tasks:

·       Performing triangulation to determine the relative distance of a point or an object that is seen in the images.

·       Rendering the Cameras in 3D.

·       Reconstructing a 3D object and finding its geometry and the true color of its surfaces.