I have a satellite image file originally created with WGS 1984. I need to use it in a drawing with British National grid co-ordinates. In Arc I redefined the projection to British grid, but when I import it still comes in using the WGS co-ordinates and is consequently in completely the wrong place. I've looked at various threads, but all seem to deal with drawing or vector data rather than raster. Any help most appreciated.
Solved! Go to Solution.
Solved by nomisc. Go to Solution.
Solved by Pointdump. Go to Solution.
Hi Antonio,
As you'll see from the attached this comes in to roughly the same area, though not exactly. What is your "insane" idea?
Cheers
Simon
Hi Dave,
I am indeed using metric, which is why I was surprised to find my default insertion units were inches, but heigh ho. We are definitely making progress though.
Cheers
simon
nomisc wrote:
What is your "insane" idea?
Well, since your original TIFF is too huge to be handled, I guess you can shrink it somehow:
1] please copy and paste the following worldfile code in Notepad (taken from your metadata), and save it as SHOTLEY.TFW
0.3164 0.0000 0.0000 -0.3164 619509.7410 238511.9190
2] open up your monster TIFF in a graphic editor (e.g. Irfanview), then decrease its colour depth to 256, finally save it as SHOTLEY.TIF with LZW compression
3] ensure that SHOTLEY.TFW and SHOTLEY.TIF lay in the same folder
4] try now to load the latter in Autocad Map: what happens?
Antonio,
"A geo-referenced image needs neither any Worldfile (because it already contains the geo-header) nor any .PRJ (concerning vector shapes only)..."
So, in this case, Shotley.tfw would overrule the geo-header in Simon's tiff?
How did you learn how to deal with all of these geo-files?
Dave
Dave Stoll
Las Vegas, Nevada
Hi,
>> So, in this case, Shotley.tfw would overrule the geo-header in Simon's tiff?
In dialog _MAPIINSERT you can select the correlation source, so it's up to the user which georef-info should be used.
- alfred -
Pointdump wrote:
So, in this case, Shotley.tfw would overrule the geo-header in Simon's tiff?
No, Dave, just because when you modify an image in a graphic editor like Irfanview, you're losing every geo-header: in this case, a worldfile is MANDATORY.
The goal of my test is the following: if the 256-color ortho will load, it will be proved the issue is caused by the colour depth.
Whereupon, we'll be able to reconstruct the suitable geo-header: stay tuned, sir.
Hi Antonio,
Again, close, but no cigar. This puts the image in the same location as I managed with Dave after changing my input units from inches to metres, but it's still not quite right.
I've had a thought and cut out a small segment of the original image as I figure if we can get that to work we can do the whole thing. It's attached so do your magic.
Thanks
Simon
Alfred,
As usual, you cut to the heart of the matter. Thank you. Is there also a place to change the "Image Correlation" in the FDO Connection?
How do you decide whether to use Data Connect or MAPIINSERT?
Dave
Dave Stoll
Las Vegas, Nevada
Antonio,
"...when you modify an image in a graphic editor like Irfanview, you're losing every geo-header..."
So that would be the subtle distinction in whether a file is called a TIFF or a GEO-TIFF.
"...stay tuned, sir."
I'm glued to this Forum. Thank you, Master Antonio!
Dave
Dave Stoll
Las Vegas, Nevada
Hi,
>> Is there also a place to change the "Image Correlation" in the FDO Connection?
No, at least I don't know an option to switch that.
In case of difficulties I would create a world-file and remove the georef-info from the image header.
>> How do you decide whether to use Data Connect or MAPIINSERT?
MAPIINSERT created referenced images that are (in most cases) faster on display. So PAN & ZOOM is faster.
...while MAPCONNECT/FDO can use coordinate systems 😉
- alfred -
Alfred,
"In case of difficulties I would create a world-file and remove the georef-info from the image header."
OK, so any image-editing will remove the geo-header, right? How do you correlate the image if the geo-reference is gone? How does a World File reference an image?
Dave
Dave Stoll
Las Vegas, Nevada
Hi,
>> so any image-editing will remove the geo-header, right?
Any image editing done with tools not knowing anything about a georef-info in the file-header, that is correct.
But there are also tools out in www that can handle image headers to define their georeference position/rotation/scaling.
>> How do you correlate the image if the geo-reference is gone?
First at all, I would like to save the georef-info and try not to lose them, so e.g. inserting the image in AutoCAD and then let RasterDesign create a world-file 😉
If I don't have any georef-info I would use e.g. RasterDesign (up to rubbersheeting) or AutoCAD command _ALIGN or Map 3D commands _ADETRANSFORM or _ADERSHEET.
>> How does a World File reference an image?
Described >>>there<<<.
- alfred -
Thank you, Alfred. I've used _ALIGN many times, but I didn't know about _ADETRANSFORM or _ADERSHEET. I also didn't know about RasterDesign. I just did some reading about it in the Help Files. I'll investigate further.
Dave
Dave Stoll
Las Vegas, Nevada
nomisc wrote:
cut out a small segment of the original image
Check the attachment out: any changes?
Please tell me: were you able to load the full color 1.5 GB imagery, or only my decreased 256-color release..?
Hi Antonio,
That works What's more it's part of the original 1.5MB file, not the 256 version. Please tell me what you did.
With reference to the whole image, I was able to load the entire file, but as discussed it was in the wrong place. However, if the segment could be loaded the same rules should work for the whole image. My only slight worry is that the image has a frame around it with no data, but hopefully that shouldn't affect it.
Thanks again for all your help.
Simon
Simon, I'm as glad as confused, because the metadata you pasted referred to metric units, while Segment.TFW contains lat/long coordinates...
Regardless of the huge 1.5 GB file stores degrees or meters, you should convert it to the following CRS:
EPSG:27700
If you don't know how to do it, ask your Gis crew, since it's not so easy as appears.
Hi Antonio and Dave,
Success!! It transpires we were very nearly there before when I converted from inches to metres. It turns out that there are a number of different transformations from WGS 1984 to OSGB (aka 27700) and the particular satellite imagery I have works best with OSGB_1936_To_WGS_1984_3 rather than the default OSGB_1936_To_WGS_1984_1! and that is what makes the subtle difference. You gave me the idea, Dave, when you mentioned the difference between US Survey Feet and International Feet so I spoke to one of our GIS guys and hey presto.
Thanks everyone for all your help and suggestions
Simon