#5 Laplacian pyramid scale wrong?

[Robert Toth]

LPyramid.cpp uses a linearly increasing filter footprint from one level to the next. The cpd array (Metric.cpp) indicates that the intent was to halve frequency content at each level, suggesting either down-sampling each level by a factor of two, or applying successively larger filters, would be appropriate. The referenced Burt Adelson '83 paper does down-sample each level.