Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - aggieair

Pages: 1 [2] 3 4 ... 7
General / Re: Single image ortho rectification
« on: December 11, 2015, 01:27:11 AM »
Based on my year experience so far, I'm going to say no because when I put down a marker/GCP, it must exist in two frames at least.  If it's in one frame, then it doesn't even get considered in the solution. (PS I work with only aerial imagery)

General / Re: Actual photo overlap and Report photo overlap - different?
« on: December 10, 2015, 11:27:06 PM »
Hi Alexey, didn't think you'd be awake!

Yes a Mesh was generated for this project (and all projects, this was before I knew about the DEM-derived option in 1.2.0).  I believe it is complete, but otherwise how is that indicated?  Used HIGH face count on a LOW dense cloud.

I did however, continue this project (after the software upgrade) with the Build DEM and Build Orthomosaic options, but it hung on to the Mesh from the original version.

UPDATE: just reran the mesh (just in case) and made another report.  Same result.

General / Re: orthomosaic coloration problems
« on: December 10, 2015, 09:43:15 PM »
Which blending mode are you using?  I typically use Mosaic all the time.

From Agisoft:

    Mosaic (default)
– images are decomposed into high frequency and low frequency components. A weighted average is calculated separately for low frequency and high frequency components (with different weights), which are subsequently combined back into final image.
    Average – simple weighted average of overlapping image colors.
    Max Intensity – color from the image with the highest intensity at the corresponding point is used (without any changes).
    Min Intensity – color from the image with the lowest intensity at the corresponding point is used (without any changes).

General / Actual photo overlap and Report photo overlap - different?
« on: December 10, 2015, 09:41:13 PM »
Normally we are used to projects with planned UAV flights and recorded flight log info.

Currently I'm on a historical aerial imagery project.  Images were scanned at 1200 dpi on a large format reflective scanner.  No flight log or orientation info was include - purely paper maps only. Markers were extracted from LiDAR information for orthorectification.

See my attached photos.  Laying the images out on the floor shows decent overlap.  The point cloud is decent and it makes a decent looking mosaic in ArcGIS.

However, in the report PDF, it looks like awful overlap and an awful mosaic.  Any reason to this?  I need to assess 24 individual projects for overlap success and was hoping to just glance at the PDF report rather than laying out hundreds of images all over again.  Agisoft projects have been started for all a generating a report is possible.  I just want to be sure its reporting it correctly.

Can anyone clarify this?

Yes, much better!  No more shards of glass issues :)

When zoomed to full extent in ArcGIS, the new DEM derived mosaics appear "softer" or not as sharp as the previous mesh-derived mosaics.  But then upon zooming in and checking the pixel size, they are nearly the same.

Sorry I have not made any dense point cloud screen shots for you yet...

I am working on comparing the new DSM rasters now derived from point clouds...

We were just about the export the points to Global Mapper and to this externally. Phew!

From Agisoft:

The issues that you are observing on the orthomosaic might be related to the low number of polygons in the area of interest. Even when you are using High face cont option, the could be areas with small polygons (related to the trees and greenery, for example) and big polygons (on the roads or bare ground).

Additionally we would like to inform you that version 1.2.0 has been released recently and it allows to generate DEM based on the dense cloud itself, so the issues the large polygons can be avoided:

This was the trick.  We were just about to create the mosaics in an external software using the LAS points as a last resort.  Now it can be done in Agisoft!

General / Re: Problem with aligning photos
« on: December 08, 2015, 10:00:59 PM »
If mine don't all align on the magic 40000,10000 starting point, then I need to fiddle with the values a lot.  I also align on high unless I'm making a quickie mosaic.  Always use reference as pair selection if your images have lat long data.

General / Analyze what is wrong with my historical aerial imagery mosaics
« on: December 08, 2015, 09:59:03 PM »
We are in the middle of a historical aerial imagery project where we scanned in paper maps at 1200 dpi.  No referenced data is known, but we pulled XYZ values from LiDAR bare earth and intensity rasters.

It's been a challenge to say the least adding the "right" arrangement of points without it blowing up. 

Anyway, some mosaics will turn out OK and others will have some issues (see attached). 

Is this mostly because I am using the standard DSM (aka all points) to create the mesh and not a DTM? It would be pretty impossible to classify ground points for a DTM with these forested scenes.  I've also been using the LOW or MEDIUM point cloud density quality so that I get a well distribution of points even up in the trees.  HIGH and ULTRA will leave lots of gaps which affects the mosaic.

General / Re: Can I change the camera file path
« on: December 08, 2015, 07:18:19 PM »
This is good to know!

General / Re: Agisoft Vs Competitors
« on: December 07, 2015, 08:51:43 PM »
I don't think Pix4D had the "classify ground" component to make a DTM as well as the DSM.  We really use both outputs, and that was the distinction for us.

General / Re: generation of dense cloud point
« on: August 13, 2015, 07:36:36 PM »
Make sure you're using Height Field in stead of Arbitrary when doing the Mesh too.  I think that helped in this one instance I showed.

General / Re: generation of dense cloud point
« on: August 13, 2015, 12:50:31 AM »
Do they look like "cliffs" where there are no actual cliffs?
What are your settings you are using?

Maybe I'll try making a terrain model only and export that model to import into the TIR.  The raggedy point cloud of the vines translate to many of the speckles.

Just talking to myself over here!

UPDATE: using a model created from ground points only did improve the speckles.

It's looking like my high density point cloud is not as clean as last year's point cloud.  I've  tried several methods to fix it up, but I think it's just what it's going to look like given my options.

Do you just accept things and move on?

Any more trouble shooting ideas?

Camera: ICI 9000 Thermal Camera
640x860 pixels
AGL: 450m
16-bit stretching

We have used this camera before for TIR flights.  However, today I'm experienced some speckles in my final mosaic but not in the raw images.  Due to the TIR not working all the way to the end in Agisoft, I am importing the RGB model from this exact day/time/flight to make my TIR mosaic.  Great workaround, but these speckles are new for us.  Last year's TIR images were not put into 16-bit but rather 15-bit I think.  So a different stretching to the TIFs this year.

I've tried exporting various blending modes, enabling color correction.  These speckles are not visible on the raw TIR images.

Any ideas?

Pages: 1 [2] 3 4 ... 7