For instance, when Type (c) and Type (d) slope weighting constraints are used, the overall normalization factors would be k w{k)ic) = - i{k - 1)] = i(K) - i{0) = N (6.68)
respectively, independent of the length of the warping functions [484]. Type (a) and type (b) slope weighting constraints will produce normalization factors that are strong functions of the actual paths, hence Type (c) and (d) are preferred in most of the applications.
Since local continuity constraints are imposed on the warping function, certain portions of the i,j grid (search space in mapping) are excluded from the region where the optimal warping path can be located. For each type of local constraints the allowable regions that determine the maximum and minimum amounts of expansion (or com-
I Smoothed )
I Smoothed )
I Smoothed*)
I Smoothed*)
Figure 6.26. Uniformly redistributed (smoothed) slope weightings for Type II local constraint [406, 484, 533].
Figure 6.26. Uniformly redistributed (smoothed) slope weightings for Type II local constraint [406, 484, 533].
pression) can be defined using the two parameters Qmax and Q
Qr max e
Lfc=i
Was this article helpful?
Post a comment