Image Rectifier: Raster Image in "Zoomed View" window GCP not being placed correctly
Jake
Global Mapper UserTrusted User
This is not a snapping issue. I have ALT clicked to no avail. Plus there is no vector data in the Zoom View window.
Basically when I try to add a GCP to the ZOOMED VIEW window the point is placed about 10 pixels away. There is no vector present. This is a raster image.
This is how I got to this point.
Loaded shape-files (lat/long).
Loaded a dumb TIF and choose to rectify it.
I used the shape-files to pick the GCPs.
Changed the projection to match the raster and saved PRJ file.
Selected triangulation as method.
Saved the GCP file (just in case).
Clicked OK in the Image Rectifier and viewed in the main window.
Saved the workspace (which I see saves the GCPs inside the GMW file).
Opened the GMW workspace next day.
Opened the Overlay Control Center and right-clicked on the TIF layer to Modify Layer Position/reprojection.
Added a few more GCPs. It was after a few iterations of this that I noticed that despite the triangulation method some GCPs seemed to be ignored as the raster and vector did not match at that point (PROBLEM 1). When I went back into the Rectifier Tool I had to UPDATE the points as they had moved off the GCPs slightly (PROBLEM 2). Also, when clicking in the Zoomed View (TIFF) the point would be offset about 10 pixels from the cursor axis (PROBLEM 3).
Note: I never exported the TIF to a file. I left it in its original form. Once I am happy with the rectification I will export it. But wondering if the constant "Modify Layer Position" is destructive or have I run into a limitation with the projection algorithm? It is Lambert Conic.
Any ideas?
Basically when I try to add a GCP to the ZOOMED VIEW window the point is placed about 10 pixels away. There is no vector present. This is a raster image.
This is how I got to this point.
Loaded shape-files (lat/long).
Loaded a dumb TIF and choose to rectify it.
I used the shape-files to pick the GCPs.
Changed the projection to match the raster and saved PRJ file.
Selected triangulation as method.
Saved the GCP file (just in case).
Clicked OK in the Image Rectifier and viewed in the main window.
Saved the workspace (which I see saves the GCPs inside the GMW file).
Opened the GMW workspace next day.
Opened the Overlay Control Center and right-clicked on the TIF layer to Modify Layer Position/reprojection.
Added a few more GCPs. It was after a few iterations of this that I noticed that despite the triangulation method some GCPs seemed to be ignored as the raster and vector did not match at that point (PROBLEM 1). When I went back into the Rectifier Tool I had to UPDATE the points as they had moved off the GCPs slightly (PROBLEM 2). Also, when clicking in the Zoomed View (TIFF) the point would be offset about 10 pixels from the cursor axis (PROBLEM 3).
Note: I never exported the TIF to a file. I left it in its original form. Once I am happy with the rectification I will export it. But wondering if the constant "Modify Layer Position" is destructive or have I run into a limitation with the projection algorithm? It is Lambert Conic.
Any ideas?
Comments
-
That is strange, what version of Global Mapper are you using? You could be seeing some old bugs with rectification that have since been fixed.
Thanks,
Mike
Global Mapper Support
support@globalmapper.com -
11.02 I think. I am not at my PC right now to verify but it was only about 1-2 months old.
-
When you get back to your PC can you first try getting the latest v11 build to see if that makes any difference? I have placed a new build at http://www.globalmapper.com/global_mapper11.zip with the latest changes for you to try. Simply download that file and extract the contents into your existing v11.xx installation folder to give it a try. If you are using the 64-bit v11 version there is a new build at http://www.globalmapper.com/global_mapper11_64bit.zip .
Thanks,
Mike
Global Mapper Support
support@globalmapper.com -
global_mapper wrote: »When you get back to your PC can you first try getting the latest v11 build to see if that makes any difference?
Starting the test.
IT WORKS NOW. This did correct all 3 issues I was having. One thing I noticed is that the triangulation now seems to be using a linear fit versus an inverse square weighted method.
A big thanks. -
linear fit versus an inverse square weighted method
Actually I take that back. The inverse square weighted method would not have produced the results I was seeing. -
The triangulated rectification should work the same as it always has, so I'm not sure why you are seeing a difference. The method is also called piecewise affine as a 3-point affine transformation is done inside each triangle.
Thanks,
Mike
Global Mapper Support
support@globalmapper.com -
Well the problems have re-surfaced. Some of the points are moving 20 pixels or about 350m on the ground).
Thanks for the extra information on the method. -
I have added a snapshot to illustrate one of the issues I am having. Method is triangulation yet GCPs do not line up after performing the triangulation. As you can see the vector and raster data does not match at the GCPs. I would expect there to be variations/tearing along the triangle edges but not at GCP. I can send you the source files if needed.
PS: Forum re-sampled from a 1.2 Mpixel 27KB PNG file to an 18KB jpeg when I uploaded it. -
PSS: Just manually checked the pixel coordinates in another software package and they are the correct values (±1 pixels). The problem is the ZOOMED VIEW plots them incorrectly. So since I am basing my positions on the slightly shifted ZOOM VIEW I am getting slightly shifted results. The shift does not seem to be consistent either.
-
Just another clarification. The image and cross-hairs in ZOOMED VIEW seem to be OK but the RED dot that is placed initially is not. As I mentioned earlier it is offset. In my attempts to move the red dot over the GCP feature I was actually moving the GCPs away from the correct location. So for now I will ignore the GCP red dot.
-
This is odd, if you can provide the data (including TIFF, Shapefile, and GMW file with control points) I can take a look and see what is happening. You can email them to support@globalmapper.com if you don't want to post here, or if they are large you can FTP to ftp.globalmapper.com with a username of 'upload' and a password of 'upload'.
Thanks,
Mike
Global Mapper Support
support@globalmapper.com -
FTP site is down. Files are quite large.
-
It is there now on hte FTP site. I was not able to give you the SHAPE files though (too many). But you can see that when you open the TIF and GCP file alone it plots properly. The GMW file may offer some clues.
-
Thanks I have the files now and can reproduce the issue. Oddly this works fine in the last official v11.02 release and in the latest debug build, but fails in the latest release build. I will investigate to see what is happening.
Thanks,
Mike
Global Mapper Support
support@globalmapper.com -
I was able to track down the issue and get it fixed. I have placed a new build at http://www.globalmapper.com/global_mapper11.zip with the change for you to try. Simply download that file and extract the contents into your existing v11.xx installation folder to give it a try. If you are using the 64-bit v11 version there is a new build at http://www.globalmapper.com/global_mapper11_64bit.zip .
Let me know if I can be of further assistance.
Thanks,
Mike
Global Mapper Support
support@globalmapper.com -
My quick reply did not appear for some reason. Anyway just to let you know the results look very promising. The GCPs are now plotting correctly and the raster and vector results coincide at the control points. It is better than I hoped. Thank you Mike for the rapid turn-around and excellent service.
-
OK one last question. Now that I am satisfied with the triangulation is there a way to apply the rectification results to lat/long vector data (points) that reside on another layer? I think not. The only way I can think of is to get the equivalent pixel coordinates. Which I don't have without some extra work.
-
You can rectify a vector layer by right-clicking on the vector layer in the Control Center and select the option to Reposition the layer. However you couldn't directly use the GCPs that you already have since, as you indicated, they refer to pixel coordinates and not coordinates in the vector layer's coordinate space.
Let me know if I can be of further assistance.
Thanks,
Mike
Global Mapper Support
support@globalmapper.com
Categories
- 12.7K All Categories
- 5.6K Features Discussion
- 342 Downloading Imagery
- 1.3K Elevation Data
- 380 Georeferencing Imagery Discussion
- 628 GM Script Language
- 53 User Scripts
- 113 GPS Features
- 414 Projection Questions
- 819 Raster Data
- 1.3K Vector Data
- 6.6K Support
- 177 Announcement and News
- 908 Bug Report
- 558 SDK
- 1.2K Suggestion Box
- 3.7K Technical Support
- 562 Other Discussion
- 129 GIS Data Sources
- 27 Global Mapper Showcase
- 233 How I use Global Mapper
- 107 Global Mapper Forum Website