Microscope Image Processing

advantage with the kNN approach is that no estimation of the pdf is required

because the function is only approximated locally, and all computation is deferred

until the classification stage. However, the disadvantages are the memory require-

ment to store training samples and the computational complexity required to

search for the k nearest samples during the classification of each unknown object.

On the other hand, with the KDE methods one can generalize the hypercube

Parzen window with a smooth nonnegative kernel function c(x) that satisfies the

condition

Ð

c(x) dx ¼ 1. Just as the Parzen window estimate can be considered

a sum of boxes centered at the samples, the smooth kernel estimate is a sum of

‘‘bumps’’ placed at the samples, and the kernel fun ...