Glint in WorldView 2 Image

8462
33
01-11-2012 06:05 PM
RamiSaeed
New Contributor
Hello All,

I got my hands on an 8 band WorldView 2 image, sadly the entire image is covered with sun glint since its a marine image... i would like to know of how to remove this glint problem... Thanks
0 Kudos
33 Replies
Luke_Pinner
MVP Regular Contributor
Sorry, I don't know. It's only been tested on 4.8/5.0 so it may not run on 4.7 especially since you had issues running the original code on 4.7.
0 Kudos
RamiSaeed
New Contributor
Hey, installed it on 4.8 and the following happened...

So i opened my file and selected my ROI for deep water:
[ATTACH=CONFIG]19008[/ATTACH]

And went ahead with the rest of the inputs:
[ATTACH=CONFIG]19009[/ATTACH]

However, when i pressed ok the following happened following with a complete software shutdown:
[ATTACH=CONFIG]19010[/ATTACH]

Any thoughts?
0 Kudos
Luke_Pinner
MVP Regular Contributor
You're trying to run it in ENVI. The script only works in ENVI+IDL.
0 Kudos
Luke_Pinner
MVP Regular Contributor
This one runs in ENVI runtime (attached)
0 Kudos
RamiSaeed
New Contributor
Hey, it worked, both of them... about the results were the first four bands where completely white, and then the 5th and 6th did show something, while the last bands where completely gray, couldn't make a composite out of the results because nothing shows... I'll fiddle with the histogram and see what i can get..
0 Kudos
Luke_Pinner
MVP Regular Contributor
I didn't bother building in any "pretty display" code.

When I run the script on radiance values (standard ENVI Worldview 2 radiance calculation) I get the following displayed:
[ATTACH=CONFIG]19135[/ATTACH]

But simply stretching the display (enhance menu) the image displays fine:
[ATTACH=CONFIG]19136[/ATTACH]

Running the script on raw digital numbers (Digitalglobe IMD format, TIFF or ENVI format) I get the image displaying ok (if a little washed out) first time.

Not a big deal, I prefer to control the display stretch myself anyway.

The band you specified as the NIR band is output as grey as the sampled data (always NIR band) is regressed against each band and when this is done for the NIR band (i.e same values), the slope will always be 1. 

Note I get better results using band 7 than band 8 for deglinting Worldview2.
0 Kudos
PaulaNieto
New Contributor
Hello!
Anyone knows how to apply this code for a Quickbird image?

Thanks!
0 Kudos
Luke_Pinner
MVP Regular Contributor
Exactly the same way. Define ROI, run the tool and follow the prompts. NIR is band 4 for Quickbird imagery.
0 Kudos
PaulaNieto
New Contributor
Thank you!
Thats what I thought, but I get this error
% Compiled module: $MAIN$.
% Attempt to subscript SAMPLES with <INT      (       4)> is out of range.
0 Kudos
PaulaNieto
New Contributor
Hello again,

About the error I have, if I change the band number to a 3 instead of a 4 (for Quickbird), the code starts, although I get another error:
M               FLOAT     =  1.00112e-005
% Unable to allocate memory: to make array.


Therefore, I guess its also a problem with band number 4.

Im new using envi, so I have no much experience. Thank you, your help is highly appreciated.
0 Kudos