Hi there -- I'm trying to use topography feature extraction to get a map of local 3 x 3 image maximum (e.g. "peaks") -- I'm noticing that "close" values (even with the slope tolerance set to 90 and curvature to 0.000) sometimes result in adjacent pixels being classified as peaks -- since a peak is, by definition, a pixel with all neighbors lower than it, and there are NOT equal values in these adjacent "peaks", I'm a bit stumped as to what might be going on.
On a related note, what, exactly, does the kernel size refer to? Considering it can take even sized kernels, I'm under the impression this value is different from those used in the texture calculations in ENVI Radar tools. Is this correct, or is the algorithm "orienting" the even sized window, for instance, slightly to the NW of the true center of the window, e.g. for a kernel size of 4, we get a window that looks like:
X X X X
X O X X
X X X X
X X X X
Where "O" is the "focal" point of the window?
Thanks!
--j
|