[GRASS5] probs with i.points

GRASS GRASS developer
1

Hi again!

I'm currently working with a series of big (my opinion) aerial photos,
each a .tiff about 380 mb (11200*11200 pixel).
Importing and viewing is working fine , with a little patience of
course.
I have 10 shoots in 2 rows , so i begun with the third in the upper row
(#3) as the origin in a x-y location.
I identified identic points in the overlap area with i.point (i.point3
won't work, it seems it forgot the identified points every time i wanted
to use the analyse-funktion, displaying something like "not enough
points marked out" although i did, anyway).
With i.rectify (using 1 degree transformation) i got nice results using
air shoot number #3 (middelst in 1st row) as reference for the direct
neighbours (#2 and #4).
Trying to use the newly georeferenced photo #2 to identify points in
number 1 i.points crushed with a message:

+------------------------------------------+------------------------------------------+

|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
+-------------------------------------------------------------------------------------+

|Use mouse now
...Speicherzugriffsfehler |

GRASS:/spare/gisdata/grass/bin

                                                         |

+---------------------------------------------------------------------------------------

I tried with downskaled tif of 90 mb - same effect.

If i used the very small overlapping strip between photo #1 and #3 to
identify points - i.rectify worked (poor results of course).
The same in the second row...

any idea?

Regards
Heiko Kehlenbrink

2

Hi Heiko,

On Mon, Oct 01, 2001 at 02:54:04PM +0200, Heiko Kehlenbrink wrote:

Hi again!

I'm currently working with a series of big (my opinion) aerial photos,
each a .tiff about 380 mb (11200*11200 pixel).
Importing and viewing is working fine , with a little patience of
course.
I have 10 shoots in 2 rows , so i begun with the third in the upper row
(#3) as the origin in a x-y location.
I identified identic points in the overlap area with i.point (i.point3
won't work, it seems it forgot the identified points every time i wanted
to use the analyse-funktion, displaying something like "not enough
points marked out" although i did, anyway).

[yes, unfortunately there are some bugs in i.point3, however, this
module is much more convenient than i.point]

With i.rectify (using 1 degree transformation) i got nice results using
air shoot number #3 (middelst in 1st row) as reference for the direct
neighbours (#2 and #4).
Trying to use the newly georeferenced photo #2 to identify points in
number 1 i.points crushed with a message:

+------------------------------------------+------------------------------------------+
| |

[...]

| |
+-------------------------------------------------------------------------------------+

|Use mouse now
...Speicherzugriffsfehler |

GRASS:/spare/gisdata/grass/bin

segmentation fault - mhhh. Could you run this again using "strace i.points"?
Perhaps this hels a bit to find the problem. Of course a debugger would
be much better, see
documents/debugging.txt
in the source code for how to debug.

I am also using i.points regularly without problems.

I tried with downskaled tif of 90 mb - same effect.

If i used the very small overlapping strip between photo #1 and #3 to
identify points - i.rectify worked (poor results of course).
The same in the second row...

Well, since you are working in an xy-location, how can your images
overlap?

Generally: It is much more recommended to use i.ortho.photo to
rectify aerial images. Only then the results will be really accurate.

Kind regards

Markus Neteler

3

Hi Markus!

Markus Neteler wrote:

Well, since you are working in an xy-location, how can your images
overlap?

I'm working in 1 xy-location but i have created 2 mapsets , let us call them import and
mosaik.
In import i've got 9 aerial images, in "mosaik" the left 1 wich becomes the origin of the
target system in
the transformation.

Generally: It is much more recommended to use i.ortho.photo to
rectify aerial images. Only then the results will be really accurate.

Yeah, i'm thinking of that too, but i don't have a dem , only a bunch of about 8 points
with gk-koordinates and heights from
wich i can identify 4 in the image.
Should i interpolate a "dem" out of them or creat a plain dummie-dem with r.mapcalc using a
mean height for example?

Best Regards

Heiko