| Re: subset an image programatically [message #46817] |
Tue, 03 January 2006 15:09 |
Jeff N.
Messages: 120
Registered: April 2005
|
Senior Member |
|
|
Aha. I would do something like this. Find the maximum and minum
values of the xpts and ypts arrays, naming them something like min_x,
max_x, min_y, max_y. If you'd like, you can add a few extra pixels so
that the vector doesn't touch the boundary of the image here too
(for example, min_x = min(xpts) -2) Then do some array subscripting on
the input image:
out_image = input_image[min_x:max_x, min_y:max_y]
Then write out_image out to a file.
Jeff
Javier Martinez wrote:
> Hi,
> thanks for the hints. I'm already try to do the job using the
> ENVI_MASK_APPLY_DOIT routine, but with this I obtain an image of the
> same dimensions than the original image but (obviously) with the pixel
> outside of the vector file masked out, and the thing that I want to do
> Its a subset image of smaller size than the original, that match the
> dimensions of the vector file. If you know a way to do this job please
> let me know.
>
> Thanks again
>
> Javier Martinez
|
|
|
|
| Re: subset an image programatically [message #46824 is a reply to message #46817] |
Tue, 03 January 2006 11:20  |
Javier Martinez
Messages: 2
Registered: January 2006
|
Junior Member |
|
|
Hi,
thanks for the hints. I'm already try to do the job using the
ENVI_MASK_APPLY_DOIT routine, but with this I obtain an image of the
same dimensions than the original image but (obviously) with the pixel
outside of the vector file masked out, and the thing that I want to do
Its a subset image of smaller size than the original, that match the
dimensions of the vector file. If you know a way to do this job please
let me know.
Thanks again
Javier Martinez
|
|
|
|
| Re: subset an image programatically [message #46826 is a reply to message #46824] |
Mon, 02 January 2006 12:05 |
Jeff N.
Messages: 120
Registered: April 2005
|
Senior Member |
|
|
I would drop the call to ENVI_GET_ROI_INFORMATION. The values of ns &
nl are set early in the code by a call to ENVI_FILE_QUERY, which is
querying the input image. So the call to ENVI_GET_ROI_INFORMATION is
both redundant and potentially a source of error if something goes
wrong in setting up the ROI's.
Also, I would set up the "black image" like this:
out_data = intarr ( ns,nl,nb) - 9999
which allows you to skip a construct like this: out_data[*,*,*]
I'm also not sure why you need to bother creating ROI's in the first
place? If all that you're doing is setting every pixel value outside
of a vector record to -9999 (or whatever value you prefer) and keeping
every value inside the vector as is, I think you have everything you
need to do this without the ROIs. The variables xf and yf (later
REFORMed to xpts and ypts) contain the pixel addresses of the original
input image that are inside the vector record, unless I am really
missing something. Why not just build a mask for the original image
using those?
Jeff
|
|
|
|
| Re: subset an image programatically [message #46831 is a reply to message #46826] |
Mon, 02 January 2006 02:44 |
lbusett
Messages: 9
Registered: March 2004
|
Junior Member |
|
|
Hi Javier,
the problem in your code seems to be that the ENVI_GET_ROI_DATA
command returns only the values of your image for the pixels that are
part of the ROI, but not their "position" in the image. If your ROI is
formed by 14000 pixels, the "roi_data" variable will be an array with
14000 elements, while the "output" that you want to write to disk is a
matrix with the same dimensions of the input image that contains data
only for the pixels that are contained in your ROI.
In order to perform your task, you have first of all to create a
"black" image with the same dimensions of the original image, then you
have to retrieve the position of each of the pixels of your ROI with
the "ENVI_GET_ROI" command, and finally you have to put the values that
you have retrieved with the ENVI_GET_ROI_DATA command in the correct
positions of the "black" image and write it to disk with the
ENVI_WRITE_ENVI_FILE command.
I think that something like this should work, at least if the images
that you want to subset are "1 band" images.
roi_data = ENVI_GET_ROI_DATA( roi_id, fid=fid, pos = [j])
ENVI_GET_ROI_INFORMATION, roi_id, ns=ns, nl=nl
roi_addresses = ENVI_GET_ROI(roi_id) ; Get the "position" of
the pixels in the ROI
out_data = intarr ( ns,nl,nb) ;
Create the "black" image"
out_data [*,*,*] = -9999 ; Assign a value to all the pixels of the
out image so that their values will not be "similar" to the values
that you extract from the ROI
out_data[roi_addresses] = roi_data ; Assign
values to the pixels of the output image that are in the ROI
ENVI_WRITE_ENVI_FILE, roi_data, ns=ns, nl=nl, $ ; Write the image
out_name='out_name', map_info=map_info
I didn't test it, so maybe it will not work properly.
If you have a single vector file that you are using to subset a series
of images with the same dimensions, i also think that you should take a
look at this thread:
http://groups.google.com/group/comp.lang.idl-pvwave/browse_t hread/thread/4b90c8f4a2c7b715/
Hope this helps,
Lorenzo
-----------------------------------------
Lorenzo Busetto
Environmental Dynamics Remote Sensing Lab.
University of Milano-Bicocca
email: lorenzo.busetto@unimib.it
tel: 00390264482848
Javier Martinez wrote:
> Hi everyone,
>
> this is my first post, I'm not a programmer and I'm trying to subset a
> series of images with vector files in evf format. Looking the envi and
> IDL help files and some code that I found in this group, I write the
> code below, and it works properly but not completely. In the final part
> where I write the envi file I obtain an image with all the roi data in
> a single line (more than 14000 samples) and I need the data in the
> original geographical shape (obviously).
>
> Thanks in advance for any help that you can give me
>
> happy new year for all you guys
>
> Javier Martinez Pincheira
> Instituto del Medio Ambiente
> Universidad de La Frontera
>
> pro batch_roi
> envi,/restore_base_save_files
> envi_batch_init, log_file='log.txt'
>
> envi_open_file, 'image', r_fid=fid
> envi_file_query, fid, ns=ns, nl=nl, nb=nb
>
> map_info = envi_get_map_info(fid=fid)
>
> evf_fname = 'vector.evf'
> evf_id = envi_evf_open(evf_fname)
>
> envi_evf_info, evf_id, num_recs=num_recs, $
> data_type=data_type, projection=projection, $
> layer_name=layer_name
>
>
>
> FOR i=0,num_recs-1 DO BEGIN
> ; read the record
> ;
> vec = ENVI_EVF_READ_RECORD(evf_id, i)
> xmap= vec(0,*)
> ymap= vec(1,*)
>
> ;just to verify
> print, 'Number of Records ' + ': ', num_recs
> print, 'Number of nodes in Record ' + $
> strtrim(i+1,2) + ': ', n_elements(vec[0,*])
>
> envi_convert_file_coordinates, fid1, xf, yf, xmap, ymap
>
>
> roi_id = ENVI_CREATE_ROI(ns=ns, nl=nl, color=4, name='shape')
>
>
> ENVI_DEFINE_ROI, roi_id, xpts=REFORM(xf, /over), $
> ypts=REFORM(yf, /over), /polygon
>
> FOR j=0,0 DO BEGIN
> ;print, roi_id
> roi_data = ENVI_GET_ROI_DATA( roi_id, fid=fid, pos = [j])
> ENVI_GET_ROI_INFORMATION, roi_id, ns=ns, nl=nl
>
> ENVI_WRITE_ENVI_FILE, roi_data, ns=ns, nl=nl,
> out_name='out_name', map_info=map_info
>
> ENDFOR
> ENDFOR
> print, roi_id
> ENVI_BATCH_EXIT
> end
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
