Representing 3D shape from video

We present an algorithm which, taking as input positions of features (corners) in every single frame of an image sequence and the sequence itself, will generate an image of a three dimensional object optimally textured. Optimally means that for each constituent triangle of the created mesh, the texture of the best resolution will be chosen. The spatial shape of the object shown in the video sequence is calculated using a factorisation method and the best texture selection is a solution proposed by the authors.