Feature request for DILATE function [message #86449] |
Sun, 10 November 2013 14:14 |
tom.grydeland
Messages: 51 Registered: September 2012
|
Member |
|
|
The DILATE function does not behave properly on the boundary of the image. This is described in the documentation:
"If edge values are important, you must pad the image with as many zeros as there are elements in the structuring element that extend off the array, in all dimensions."
What is the purpose of this? If edge values aren't important, I can discard them. When I ask IDL to compute a value for me, it is because I care about that value.
Yes, I have completed a version of DILATE which does the padding for me, but -- please -- that shouldn't be necessary!
I'd like to request a future version of IDL to compute valid values all the way to the edge. For those who want to preserve the current behaviour, you can add a keyword /BOGUS_EDGES which restores today's behaviour. Or, if you for some unfathomable reason should decide to default to the current behaviour, you could add a keyword /CORRECT_EDGES.
Thank you in advance, and best regards,
--T (Tom Grydeland)
|
|
|