Globally Optimal Image Mosaics
Kirk L. Duffin, William A. Barrett · 1998
In this paper we examine the simultaneous solution of a set of image transformations with the goal of creating a globally optimal image mosaic. We examine an alternative parameterization of the full projective matrix transformation that leads to elimination of independent skew and aspect ratio parameters for each image. We also create a scale-free distance error metric which prevents the tendency of simultaneously solved systems to tend toward the zero solution.