In this thesis we tackle the problem of 3DS conversion, i.e. the process of transforming 2D images to a 3D Stereo form. Our contributions go in the direction of building a complete pipeline for automatic 3DS conversion of uncalibrated images.
Given the nature of the problem, i.e. receiving video sequences as input, all the building blocks share the idea that the proposed methods should be able to gather and exploit information coming from several images.
We present evidence of our contributions gathering all the results that have been presented in peer reviewed publications and cast them in a unified manner.
