PDF version of this document

Improved Model-based Objective Function

As already adumbrated in the text, towards the end of April 2004, many solutions were found which substantially improved the objective function and finally made it work. Further options also made it work relatively efficiently and obtain impressive results. Details on the changes which were applied will be explained later in this section.

Figure below illustrates how registration practically operates upon the data. It can be observed that, in this case, while the model-based objective function guides transformation, data bumps align increasingly better. Therfore, good registration is finally achieved, driven purely by model complexity.

Figure: Data being visualised by AART. In this case, 5 bumps are shown at some arbitrary state during registration.

See this with reference to a previous figure that was shown on page . The model-based algorithm itself will be explained in Section .