Alignment and Data Registration

David F. Wiley

A commonly occurring problem where two or more data sets must be aligned to one another for the purpose of comparison, blending, visualization, and analysis. Rotating, scaling, and translating the 2D or 3D data sets usually accomplish this so that the each correspond to one another based upon some specified error metric.

Non-linear alignment and registration is needed for more data where the reconstruction process or some other inherent problem causes the data itself to deform, for example, taking a retinal scan from a human eye, the resulting data is captured in a rectangular shape but instead represents the curved surface of the retina. Non-linear methods are required to deform the data into its appropriate shape.