Forum

Author Topic: Again: Detecting Coded Targets  (Read 3488 times)

Roehrich

  • Newbie
  • *
  • Posts: 5
    • View Profile
Again: Detecting Coded Targets
« on: January 29, 2018, 03:30:35 PM »
My first post seems to be broken because of the attachments...

I'm wondering why Photoscan (1.4.0) detects all coded targets in my chunk but on each individual picture only one ore two?
Can someone explain this to me?
Thanks a lot in advance!
Roehrich

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14110
    • View Profile
Re: Again: Detecting Coded Targets
« Reply #1 on: January 29, 2018, 04:01:07 PM »
Hello Roehrich,

The targets may be too large (the size of the target in pixels on the individual photo), so only smaller targets in the background are detected.
Best regards,
Alexey Pasumansky,
Agisoft LLC

Roehrich

  • Newbie
  • *
  • Posts: 5
    • View Profile
Re: Again: Detecting Coded Targets
« Reply #2 on: January 29, 2018, 04:32:53 PM »
Hi,

thank you for the comment. Is there a known limit for the size or a recommended range for the the size of the targets? In earlier versions of photscan this seemed to be different - there the same size got detected.

Roehrich

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14110
    • View Profile
Re: Again: Detecting Coded Targets
« Reply #3 on: January 29, 2018, 05:05:17 PM »
Hello Roehrich,

It is recommended to keep the size of the central black circle in the middle of the target with the radius lower than 30 pixels.
Best regards,
Alexey Pasumansky,
Agisoft LLC

Mario

  • Newbie
  • *
  • Posts: 3
    • View Profile
Re: Again: Detecting Coded Targets
« Reply #4 on: July 19, 2018, 06:16:38 PM »
And what would be the minimum size of the central black circle in the middle of the target in pixels? I want to use them in an aerial survey with 1-3 cm GSD if possible. And how to print them from PhotoScan in a pdf file on A3, A2 or custom paper size?

Dave Martin

  • Full Member
  • ***
  • Posts: 170
    • View Profile
Re: Again: Detecting Coded Targets
« Reply #5 on: July 19, 2018, 10:58:40 PM »
........ And how to print them from PhotoScan in a pdf file on A3, A2 or custom paper size?

Mario,

From PhotoScan, print them to a PDF (using e.g Primo PDF); then open the PDF using free Acrobat Reader and print from the PDF to your physical printer; when you print from Acrobat you can set the Zoom ratio from a small fraction to manifold magnification.

Dave

Guillermo Castilla

  • Newbie
  • *
  • Posts: 3
    • View Profile
Re: Again: Detecting Coded Targets
« Reply #6 on: July 20, 2018, 08:39:50 PM »
So the central black circle in the CT should not exceed 30 pixels in diameter, but what  would be the minimum diameter in pixels? If you can see the attachment I included in this post, the central black circle in this case (12 bit CT #3) is 5 pixels in diameter, but was not detected (and neither were two other CTs we deployed in this UAV test) in any of the dozens of images where it was visible. Alex, is it too small, are there other problems, or what is your recommendation?

wizprod

  • Jr. Member
  • **
  • Posts: 79
    • View Profile
    • Dronographica
Re: Again: Detecting Coded Targets
« Reply #7 on: July 20, 2018, 10:57:31 PM »
I would say 5 px is too small.
You have to use your GSD to calculate the optimum size. You could download the Pix4D GSD calculator excel sheet, and modify it slightly.
I find that printing the targets of A3's (edge to edge) will work up 25 meters AGL with a Phantom 4 Pro. That can give you some kind of reference as well.

Guillermo Castilla

  • Newbie
  • *
  • Posts: 3
    • View Profile
Re: Again: Detecting Coded Targets
« Reply #8 on: July 20, 2018, 11:15:33 PM »
Thanks widproz.

We intend to send a number of CT targets to a signage company for printing in 2 by 2 feet (60cm) white plastic squares, so that they can be reused for different missions instead of printing them in paper each time. We usually work with GSDs from 5 cm to a few mm, so it looks like that the automated CT detection would only work for a narrower range using the the 2'x2' targets (15 cm diameter for the central circle). If Alexey could confirm what the minimum diameter in pixels of the central black circle can be, we could estimate that range...