PDF version of this document

Model-based Objective Function

A valid prototype for this registration method was already in place at the start. The way this function operates has been explained in earlier parts of the report. Figure shows how measures of MSD, Generalisation ability and specificity change at each function evaluation step. The changes are expressed per cent with respect to the evaluation taken for the previous iteration. It appears rather noisy and reflects on the bad forms of this function before it was revised. As most lines remain flat, it is shown that values go in no particular direction^7.12.

Figure: Various measures shown as the old model-based algorithm proceeds.

Due to this poor performance, rigourous work began to obliterate known issues and weaknesses, resulting in an improved model-based objective function.