| IDL/ENVI SPOT-5 Level 1a (DIMAP format) - simple(?) gain problem [message #72214] |
Sun, 22 August 2010 09:31  |
D2
Messages: 3
Registered: August 2010
|
Junior Member |
|
|
Hi all,
It's a Sunday afternoon and I'm pulling my hair out trying to do
something I thought was going to be relatively easy, but instead has
turned into a bit of a nightmare for a beginner at IDL/ENVI.
I've got a geotiff of SPOT-5 level 1a imagery that I bring in by
opening up the supplied *.DIM file so that I get most of the important
metadata (e.g., band wavelengths, standard gains and offsets, etc).
So, the question is this: how can I apply separate gains and offsets
to each of my 4-bands, and then write those new results together into
a separate file, or even the original file? In other words, the
original file + 4 new "bands" added to it, or new file (with all
projection metadata, etc. intact) with 4 new band).
The original file is in 8bit "digital number" (aren't they all
"digital"?? heh.) and I'm calculating 32-bit floating point values, so
it's your standard "radiometric calibration" process.
I've tried using the "apply gain and offsets" module, but that
multiples when I want to divide values. I've also tried using band
math, but I seem to be only able to apply one equation to one band at
a time, and then it only outputs the results to a single file. I'd
like to apply 4 separate equations to their respective bands, and then
output these new results to a single file, or just write them into the
original file.
There's a user submitted code on the ENVI user forum that basically
does what I need to do (calibrate_spot.sav), but I'd like to get under
the hood to tweak parameters and begin to teach myself to understand
IDL. What I'm looking to do is apply my own custom gains and offsets
instead of using those supplied in the metadata file.
I do want to learn, and I'm eagerly awaiting my Morton Canty book to
arrive, but I'd like to make some headway on this now.
Any pointers (and sample code) would be immensely appreciated!
Cheers,
Dennis
|
|
|
|
| Re: IDL/ENVI SPOT-5 Level 1a (DIMAP format) - simple(?) gain problem [message #72305 is a reply to message #72214] |
Mon, 23 August 2010 15:16  |
D2
Messages: 3
Registered: August 2010
|
Junior Member |
|
|
On Aug 23, 4:58 pm, Maxwell Peck <maxjp...@gmail.com> wrote:
> On Aug 24, 4:44 am, D2 <dennis.d...@gmail.com> wrote:
>
>
>
>> On Aug 23, 12:53 pm, "Jeff N." <jeffnettles4...@gmail.com> wrote:
>
>>> On Aug 22, 5:15 pm, Maxwell Peck <maxjp...@gmail.com> wrote:
>
>>>> On Aug 23, 2:31 am, D2 <dennis.d...@gmail.com> wrote:
>
>>>> > Hi all,
>
>>>> > It's a Sunday afternoon and I'm pulling my hair out trying to do
>>>> > something I thought was going to be relatively easy, but instead has
>>>> > turned into a bit of a nightmare for a beginner at IDL/ENVI.
>
>>>> > I've got a geotiff of SPOT-5 level 1a imagery that I bring in by
>>>> > opening up the supplied *.DIM file so that I get most of the important
>>>> > metadata (e.g., band wavelengths, standard gains and offsets, etc).
>
>>>> > So, the question is this: how can I apply separate gains and offsets
>>>> > to each of my 4-bands, and then write those new results together into
>>>> > a separate file, or even the original file? In other words, the
>>>> > original file + 4 new "bands" added to it, or new file (with all
>>>> > projection metadata, etc. intact) with 4 new band).
>
>>>> > The original file is in 8bit "digital number" (aren't they all
>>>> > "digital"?? heh.) and I'm calculating 32-bit floating point values, so
>>>> > it's your standard "radiometric calibration" process.
>
>>>> > I've tried using the "apply gain and offsets" module, but that
>>>> > multiples when I want to divide values. I've also tried using band
>>>> > math, but I seem to be only able to apply one equation to one band at
>>>> > a time, and then it only outputs the results to a single file. I'd
>>>> > like to apply 4 separate equations to their respective bands, and then
>>>> > output these new results to a single file, or just write them into the
>>>> > original file.
>
>>>> > There's a user submitted code on the ENVI user forum that basically
>>>> > does what I need to do (calibrate_spot.sav), but I'd like to get under
>>>> > the hood to tweak parameters and begin to teach myself to understand
>>>> > IDL. What I'm looking to do is apply my own custom gains and offsets
>>>> > instead of using those supplied in the metadata file.
>
>>>> > I do want to learn, and I'm eagerly awaiting my Morton Canty book to
>>>> > arrive, but I'd like to make some headway on this now.
>
>>>> > Any pointers (and sample code) would be immensely appreciated!
>
>>>> > Cheers,
>
>>>> > Dennis
>
>>>> From memory the gain/offset routine applies them in the 'remote
>>>> sensing' sense, that is GAIN* (DN - OFFSET) . If you adjust your
>>>> offset accordingly the routines will do what you want.
>
>>> I've used envi's gain/offset very little, and a long time ago at that,
>>> so i could be wrong but i think that that routine applies the same
>>> gain/offset to every band, and the OP mentioned a different gain/
>>> offset for every band.
>
>>> Assuming that's the case, what I would do is assemble all the gain's
>>> into a "spectrum" (a linear array with one element per band containing
>>> all the gains), and then do the same thing for the offsets. From that
>>> point what you'd want to do is actually spectral math:
>
>>> s1*s2+s3 ;adjust equation as you need to
>
>>> where:
>
>>> s1 - input file (use "Map variable to input file" button)
>>> s2 - gain "spectrum"
>>> s3 - offset "spectrum"
>
>>> Of course, if you were going to be doing this several times, and have
>>> ENVI+IDL, i'd just write code to do this.
>
>>> Jeff
>
>> Thanks Jeff. Much appreciated.
>
>> Originally, I wanted to use the the gain/offset module as it outputs
>> the calculated bands into a single file, but I don't think there's a
>> way of reusing it to meet my needs. It calculates TOA radiance (and
>> applies separate gains and offsets to each band) like so:
>
>> TOA Radiance = (DN*GAIN)-offset
>
>> but I need:
>
>> TOA Radiance = (DN/GAIN)-offset
>
>> However, your solution does provide me with an option, so thanks for
>> that. I'd like to program it all in IDL, but I'm a complete beginner.
>> Thankfully, my "Practical IDL Programming" book has just arrived, so
>> hopefully there will be some tips in there.
>
>> Let me know if you have any suggestions/caveats about how to use ENVI
>> specific functions to accomplish this (I'm still waiting for my Morton
>> Canty ENVI//IDL book to arrive).
>
>> Thanks again!
>
>> Dennis
>
> You can definitely use the gain/offsets differently on each band. 1/
> GAIN might be useful ??
So elegant (and obvious), I didn't pick it up the first time you
suggested it. Thanks Maxwell!!
|
|
|
|
| Re: IDL/ENVI SPOT-5 Level 1a (DIMAP format) - simple(?) gain problem [message #72307 is a reply to message #72214] |
Mon, 23 August 2010 13:58 |
Maxwell Peck
Messages: 61
Registered: February 2010
|
Member |
|
|
On Aug 24, 4:44 am, D2 <dennis.d...@gmail.com> wrote:
> On Aug 23, 12:53 pm, "Jeff N." <jeffnettles4...@gmail.com> wrote:
>
>
>
>> On Aug 22, 5:15 pm, Maxwell Peck <maxjp...@gmail.com> wrote:
>
>>> On Aug 23, 2:31 am, D2 <dennis.d...@gmail.com> wrote:
>
>>>> Hi all,
>
>>>> It's a Sunday afternoon and I'm pulling my hair out trying to do
>>>> something I thought was going to be relatively easy, but instead has
>>>> turned into a bit of a nightmare for a beginner at IDL/ENVI.
>
>>>> I've got a geotiff of SPOT-5 level 1a imagery that I bring in by
>>>> opening up the supplied *.DIM file so that I get most of the important
>>>> metadata (e.g., band wavelengths, standard gains and offsets, etc).
>
>>>> So, the question is this: how can I apply separate gains and offsets
>>>> to each of my 4-bands, and then write those new results together into
>>>> a separate file, or even the original file? In other words, the
>>>> original file + 4 new "bands" added to it, or new file (with all
>>>> projection metadata, etc. intact) with 4 new band).
>
>>>> The original file is in 8bit "digital number" (aren't they all
>>>> "digital"?? heh.) and I'm calculating 32-bit floating point values, so
>>>> it's your standard "radiometric calibration" process.
>
>>>> I've tried using the "apply gain and offsets" module, but that
>>>> multiples when I want to divide values. I've also tried using band
>>>> math, but I seem to be only able to apply one equation to one band at
>>>> a time, and then it only outputs the results to a single file. I'd
>>>> like to apply 4 separate equations to their respective bands, and then
>>>> output these new results to a single file, or just write them into the
>>>> original file.
>
>>>> There's a user submitted code on the ENVI user forum that basically
>>>> does what I need to do (calibrate_spot.sav), but I'd like to get under
>>>> the hood to tweak parameters and begin to teach myself to understand
>>>> IDL. What I'm looking to do is apply my own custom gains and offsets
>>>> instead of using those supplied in the metadata file.
>
>>>> I do want to learn, and I'm eagerly awaiting my Morton Canty book to
>>>> arrive, but I'd like to make some headway on this now.
>
>>>> Any pointers (and sample code) would be immensely appreciated!
>
>>>> Cheers,
>
>>>> Dennis
>
>>> From memory the gain/offset routine applies them in the 'remote
>>> sensing' sense, that is GAIN* (DN - OFFSET) . If you adjust your
>>> offset accordingly the routines will do what you want.
>
>> I've used envi's gain/offset very little, and a long time ago at that,
>> so i could be wrong but i think that that routine applies the same
>> gain/offset to every band, and the OP mentioned a different gain/
>> offset for every band.
>
>> Assuming that's the case, what I would do is assemble all the gain's
>> into a "spectrum" (a linear array with one element per band containing
>> all the gains), and then do the same thing for the offsets. From that
>> point what you'd want to do is actually spectral math:
>
>> s1*s2+s3 ;adjust equation as you need to
>
>> where:
>
>> s1 - input file (use "Map variable to input file" button)
>> s2 - gain "spectrum"
>> s3 - offset "spectrum"
>
>> Of course, if you were going to be doing this several times, and have
>> ENVI+IDL, i'd just write code to do this.
>
>> Jeff
>
> Thanks Jeff. Much appreciated.
>
> Originally, I wanted to use the the gain/offset module as it outputs
> the calculated bands into a single file, but I don't think there's a
> way of reusing it to meet my needs. It calculates TOA radiance (and
> applies separate gains and offsets to each band) like so:
>
> TOA Radiance = (DN*GAIN)-offset
>
> but I need:
>
> TOA Radiance = (DN/GAIN)-offset
>
> However, your solution does provide me with an option, so thanks for
> that. I'd like to program it all in IDL, but I'm a complete beginner.
> Thankfully, my "Practical IDL Programming" book has just arrived, so
> hopefully there will be some tips in there.
>
> Let me know if you have any suggestions/caveats about how to use ENVI
> specific functions to accomplish this (I'm still waiting for my Morton
> Canty ENVI//IDL book to arrive).
>
> Thanks again!
>
> Dennis
You can definitely use the gain/offsets differently on each band. 1/
GAIN might be useful ??
|
|
|
|
| Re: IDL/ENVI SPOT-5 Level 1a (DIMAP format) - simple(?) gain problem [message #72308 is a reply to message #72214] |
Mon, 23 August 2010 11:44 |
D2
Messages: 3
Registered: August 2010
|
Junior Member |
|
|
On Aug 23, 12:53 pm, "Jeff N." <jeffnettles4...@gmail.com> wrote:
> On Aug 22, 5:15 pm, Maxwell Peck <maxjp...@gmail.com> wrote:
>
>
>
>> On Aug 23, 2:31 am, D2 <dennis.d...@gmail.com> wrote:
>
>>> Hi all,
>
>>> It's a Sunday afternoon and I'm pulling my hair out trying to do
>>> something I thought was going to be relatively easy, but instead has
>>> turned into a bit of a nightmare for a beginner at IDL/ENVI.
>
>>> I've got a geotiff of SPOT-5 level 1a imagery that I bring in by
>>> opening up the supplied *.DIM file so that I get most of the important
>>> metadata (e.g., band wavelengths, standard gains and offsets, etc).
>
>>> So, the question is this: how can I apply separate gains and offsets
>>> to each of my 4-bands, and then write those new results together into
>>> a separate file, or even the original file? In other words, the
>>> original file + 4 new "bands" added to it, or new file (with all
>>> projection metadata, etc. intact) with 4 new band).
>
>>> The original file is in 8bit "digital number" (aren't they all
>>> "digital"?? heh.) and I'm calculating 32-bit floating point values, so
>>> it's your standard "radiometric calibration" process.
>
>>> I've tried using the "apply gain and offsets" module, but that
>>> multiples when I want to divide values. I've also tried using band
>>> math, but I seem to be only able to apply one equation to one band at
>>> a time, and then it only outputs the results to a single file. I'd
>>> like to apply 4 separate equations to their respective bands, and then
>>> output these new results to a single file, or just write them into the
>>> original file.
>
>>> There's a user submitted code on the ENVI user forum that basically
>>> does what I need to do (calibrate_spot.sav), but I'd like to get under
>>> the hood to tweak parameters and begin to teach myself to understand
>>> IDL. What I'm looking to do is apply my own custom gains and offsets
>>> instead of using those supplied in the metadata file.
>
>>> I do want to learn, and I'm eagerly awaiting my Morton Canty book to
>>> arrive, but I'd like to make some headway on this now.
>
>>> Any pointers (and sample code) would be immensely appreciated!
>
>>> Cheers,
>
>>> Dennis
>
>> From memory the gain/offset routine applies them in the 'remote
>> sensing' sense, that is GAIN* (DN - OFFSET) . If you adjust your
>> offset accordingly the routines will do what you want.
>
> I've used envi's gain/offset very little, and a long time ago at that,
> so i could be wrong but i think that that routine applies the same
> gain/offset to every band, and the OP mentioned a different gain/
> offset for every band.
>
> Assuming that's the case, what I would do is assemble all the gain's
> into a "spectrum" (a linear array with one element per band containing
> all the gains), and then do the same thing for the offsets. From that
> point what you'd want to do is actually spectral math:
>
> s1*s2+s3 ;adjust equation as you need to
>
> where:
>
> s1 - input file (use "Map variable to input file" button)
> s2 - gain "spectrum"
> s3 - offset "spectrum"
>
> Of course, if you were going to be doing this several times, and have
> ENVI+IDL, i'd just write code to do this.
>
> Jeff
Thanks Jeff. Much appreciated.
Originally, I wanted to use the the gain/offset module as it outputs
the calculated bands into a single file, but I don't think there's a
way of reusing it to meet my needs. It calculates TOA radiance (and
applies separate gains and offsets to each band) like so:
TOA Radiance = (DN*GAIN)-offset
but I need:
TOA Radiance = (DN/GAIN)-offset
However, your solution does provide me with an option, so thanks for
that. I'd like to program it all in IDL, but I'm a complete beginner.
Thankfully, my "Practical IDL Programming" book has just arrived, so
hopefully there will be some tips in there.
Let me know if you have any suggestions/caveats about how to use ENVI
specific functions to accomplish this (I'm still waiting for my Morton
Canty ENVI//IDL book to arrive).
Thanks again!
Dennis
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
