| What could cause disappearing array? [message #69482] |
Tue, 19 January 2010 13:54  |
robintw
Messages: 37
Registered: March 2009
|
Member |
|
|
Hi,
I've got a very strange problem in one of my IDL programs. I have two
nested FOR loops with some processing happening in the inside loop, and
after a bit of processing one of my arrays seems to disappear.
I've checked this with some strategically placed help statements, when
it seems to disappear it there (comes up with the right dimensions etc
with the help command) at the bottom of the loop, but by the time the
loop starts again it is showing as undefined.
I was about to tell you how many iterations of my loop this occurred
after (by looking at the value of i and j when it crashed) but after
it's crashed I find i and j to both be equal to 0 - even though it's
gone through most of the loop already.
I can't see anywhere in the code that I'm assigning anything to i or j.
I've left the FOR loop to look after them itself, yet somehow they end
up as 0.
When I comment out all of the code inside the for loop then the loop
runs perfectly to the end, and i and j don't get reset to zero part way
through.
Does anyone have any ideas what on earth might be causing this? I've
really got no idea!
The code for this function is below. If you need any of my code from
other functions called by this function then just let me know.
Cheers,
Robin
PRO TRY_PLACING_REGION, WholeImage, edge_map, window_size
COMMON SegData, seg_id, segment_image
d = (window_size - 1) / 2
image_dims = SIZE(WholeImage, /DIMENSIONS)
ns = image_dims[0]
nl = image_dims[1]
FOR i = 0L, nl - 1 DO BEGIN
FOR j = 0L, ns - 1 DO BEGIN
print, i
print, j
help, segment_image
j_lower_limit = j-d
IF j_lower_limit LT 0 THEN j_lower_limit = 0
j_upper_limit = j+d
IF j_upper_limit GT ns - 1 THEN j_upper_limit = ns - 1
i_lower_limit = i-d
IF i_lower_limit LT 0 THEN i_lower_limit = 0
i_upper_limit = i+d
IF i_upper_limit GT nl - 1 THEN i_upper_limit = nl - 1
image_moving_window = WholeImage[j_lower_limit:j_upper_limit,
i_lower_limit:i_upper_limit, *]
edge_moving_window = edge_map[j_lower_limit:j_upper_limit,
i_lower_limit:i_upper_limit]
segment_moving_window =
segment_image[j_lower_limit:j_upper_limit, i_lower_limit:i_upper_limit]
; Go to next possibility if moving window contains edges
if total(edge_moving_window) NE 0 THEN CONTINUE
; If any of the window's pixels are already in a segment then go
to next possibility
if total(segment_moving_window) NE 0 THEN CONTINUE
inequality_result = INEQUALITY_FUNCTION(image_moving_window, [1,
1, 1])
; If one of the pixels had a inequality value greater than 1 then
continue
if inequality_result EQ 0 THEN CONTINUE
; If we've got to here then it's a suitable seed
print, "Found a suitable seed"
; Mark the seed as a segment
segment_image[j_lower_limit:j_upper_limit,
i_lower_limit:i_upper_limit] = seg_id
help, segment_image
seg_id++
ENDFOR
ENDFOR
ENVI_ENTER_DATA, segment_image
END
|
|
|
|
| Re: What could cause disappearing array? [message #69533 is a reply to message #69482] |
Fri, 22 January 2010 13:00  |
robintw
Messages: 37
Registered: March 2009
|
Member |
|
|
On 21/01/2010 23:17, Karl wrote:
>
> If ENVI_ENTER_DATA clobbers that parm somehow, then that would explain
> why a single call to TRY_PLACING_REGION works, but more than one call
> does not.
Well done Karl! That's the problem. I had managed to make it work
earlier today when I did various alterations, including moving the
ENVI_ENTER_DATA call, but now I know why it works.
Thank you very much,
Robin
|
|
|
|
| Re: What could cause disappearing array? [message #69545 is a reply to message #69482] |
Thu, 21 January 2010 15:17 |
Karl[1]
Messages: 79
Registered: October 2005
|
Member |
|
|
On Jan 21, 4:08 pm, Mark <mark.h...@gmail.com> wrote:
> On Jan 20, 9:42 pm, Robin Wilson <r.t.wil...@rmplc.co.uk> wrote:
> ...
>
>> FOR win_size = 5, 8 DO BEGIN
>> TRY_PLACING_REGION, WholeImage, edge_map, 8
>> ENDFOR
>
> I'm curious about that code. As far as I can see, the loop index,
> win_size, is not used inside the loop, so this code has the effect of
> calling TRY_PLACING_REGION four times with the same arguments. Is that
> what you intended?
Good point, but not relevant.
What does ENVI_ENTER_DATA do with your segment_image parm?
Try checking the status of segment_image before and after the call to
ENVI_ENTER_DATA.
If ENVI_ENTER_DATA clobbers that parm somehow, then that would explain
why a single call to TRY_PLACING_REGION works, but more than one call
does not.
Perhaps you want:
FOR win_size = 5, 8 DO BEGIN
segment_image = lonarr(dims[2] + 1, dims[4] + 1)
TRY_PLACING_REGION, WholeImage, edge_map, win_size
ENDFOR
assuming that you want a fresh segment_image (zero filled) for each
iteration.
|
|
|
|
| Re: What could cause disappearing array? [message #69547 is a reply to message #69482] |
Thu, 21 January 2010 12:54 |
robintw
Messages: 37
Registered: March 2009
|
Member |
|
|
> After you crash, does IDL> help,/traceback show you to be
> where you think you are?
Hi,
Thanks for the idea. Sadly, when I call help, /traceback it shows me to
be exactly where I expect it to be.
Cheers,
Robin
|
|
|
|
| Re: What could cause disappearing array? [message #69553 is a reply to message #69482] |
Thu, 21 January 2010 10:15 |
wlandsman
Messages: 743
Registered: June 2000
|
Senior Member |
|
|
On Jan 21, 12:33 pm, Robin Wilson <r.t.wil...@rmplc.co.uk> wrote:
> On 21/01/2010 17:17, Ed Hyer wrote:
>
> Thanks for the idea. I've renamed all of the loop variables I'm using so
> they have descriptive names (such as row, column) rather than just i, j,
> k etc. However, I still get the array disappearing!
>
> Anyone got any other ideas?
A longshot but the usual reason for a disappearing array is that you
are not where you think you are, i.e. the program is crashing inside a
different procedure/function (probably with similar variable
names). After you crash, does IDL> help,/traceback show you to be
where you think you are?
|
|
|
|
| Re: What could cause disappearing array? [message #69557 is a reply to message #69482] |
Thu, 21 January 2010 09:33 |
robintw
Messages: 37
Registered: March 2009
|
Member |
|
|
On 21/01/2010 17:17, Ed Hyer wrote:
>>> "R.G. Stockwell"<noem...@please.com> wrote in message
>>>> One note: that array is in a common block, who knows who else is
>>>> manipulating it.
>
> Change the loop variables, see what happens.
Thanks for the idea. I've renamed all of the loop variables I'm using so
they have descriptive names (such as row, column) rather than just i, j,
k etc. However, I still get the array disappearing!
Anyone got any other ideas?
Robin
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
