| LABEL_REGION [message #5576] |
Tue, 09 January 1996 00:00  |
Hermann Mannstein
Messages: 22
Registered: September 1995
|
Junior Member |
|
|
Hello,
LABEL_REGION (IDL 4.01) has a nasty behaviour :
IDL> d=361
IDL> a=(indgen(d,d) mod 2)^2
IDL> b=a*rotate(a,1) ; produce a dot-pattern
IDL> a=label_region(b)
Bus error(coredump)
if the number of regions exceeds 2^15(or something like this),
it ends the IDL-session ungracefully. With d=363 I got a
-- Memory fault(coredump) (on a SGI), with d=361 a
Segmentation fault(coredump). Is there a version around,
which allows for more regions - i. e. switches to long integer if
necessary? or at least catches the error?
Thanks in advance
Regards,
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~
Hermann Mannstein Tel.: +49 8153 28-2503
Institut fuer Physik der Atmosphaere or -2558
DLR - Oberpfaffenhofen Fax.: +49 8153 28-1841
Postfach 1116 \ mailto:H.Mannstein@dlr.de
D-82230 Wessling \ 0 http://www.op.dlr.de/~pa64
Germany ________\/|________
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~`--------\--------'~ ~~~~~~~~~~~~~~~~~~~~
\
|
|
|
|
| Re: Label_Region [message #23042 is a reply to message #5576] |
Fri, 29 December 2000 12:05  |
Pavel A. Romashkin
Messages: 531
Registered: November 2000
|
Senior Member |
|
|
Nice thread, Ben :-)
This is sort of newsgroup posting I,ve been doing lately.
Cheers,
Pavel
Ben Tupper wrote:
>
> Hello,
>
> When the label region is used to uniquely mark blobs within an image,
> the edge pixels are all assumed to be zero. I would like to preserve
> those edge pixels AND perform the region labeling.
>
> I can imagine some brute-force-and-ignorance schemes to perform the blob
> coloring on my own,
> but I would like to be smart about it. Any suggestions?
>
> Thanks,
>
> Ben>
|
|
|
|
| Re: Label_Region [message #23043 is a reply to message #5576] |
Fri, 29 December 2000 11:00 |
btt
Messages: 345
Registered: December 2000
|
Senior Member |
|
|
Ben Tupper wrote:
>
> my associate-non-programmer-biologist-lunchpal suggested padding the
> image with a 1-pixel border. Of course it works like a charm.
>
He gets the prize for thinking outside the box, so to speak.
FYI, the documentation for LABEL_REGION(data) states that data can be an
N-dimensional array (at least in IDL5.4 where the EIGHT keyword has been
succeeded by the ALL_NEIGHBORS keyword.) However, while building a
wrapper around the function, I discovered that N cannot be 1.
Ben
--
Ben Tupper
Bigelow Laboratory for Ocean Sciences
180 McKown Point Rd.
W. Boothbay Harbor, ME 04575
btupper@bigelow.org
|
|
|
|
| Re: Label_Region [message #23050 is a reply to message #5576] |
Wed, 27 December 2000 13:36 |
btt
Messages: 345
Registered: December 2000
|
Senior Member |
|
|
Well... I don't mind being a fool, I just dislike making such a public
display of it.
After dutifully wrestling with all kinds of beautiful, thoughtful and
insightful uses of SEARCH2d and other goodies... none of which worked...
my associate-non-programmer-biologist-lunchpal suggested padding the
image with a 1-pixel border. Of course it works like a charm.
Sasafrasabagarashta...
Ben
Ben Tupper wrote:
>
> Hello,
>
> When the label region is used to uniquely mark blobs within an image,
> the edge pixels are all assumed to be zero. I would like to preserve
> those edge pixels AND perform the region labeling.
>
> I can imagine some brute-force-and-ignorance schemes to perform the blob
> coloring on my own,
> but I would like to be smart about it. Any suggestions?
>
> Thanks,
>
> Ben
> --
> Ben Tupper
> Bigelow Laboratory for Ocean Sciences
> 180 McKown Point Rd.
> W. Boothbay Harbor, ME 04575
> btupper@bigelow.org
--
Ben Tupper
Bigelow Laboratory for Ocean Sciences
180 McKown Point Rd.
W. Boothbay Harbor, ME 04575
btupper@bigelow.org
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
