Forum

Author Topic: The criterion used in Key Point Limit  (Read 4914 times)

Ryuseiken

  • Newbie
  • *
  • Posts: 21
    • View Profile
The criterion used in Key Point Limit
« on: November 12, 2019, 03:21:02 PM »
I have recently recognized that a very small "Key point limit" coupled with image downscaling (low alignment accuracy setting) sometimes greatly improves SfM accuracy.

But I don't find any information about what kind of criteria Metashape uses to select the features up to "Key point limit."
For example, the open source SfM software COLMAP selects the large scale features up to "max_num_features" setting.
Is Metashape using a similar criterion or combining multiple criteria including scale and distinctness?
If it is not a company secret, I'd like to know.

The situation I recognized the importance of  "Key point limit" is as follows.
For a project of 664 Phantom 4 RTK images (nadir + oblique) from 110 m altitude,  I got the following total RMS error values for 23 check points:

Case 1. Alignment accuracy "High" and Key point limit 50000: 0.8514 [m]
Case 2. Alignment accuracy "High" and Key point limit 1000:  0.4446 [m]
Case 3. Alignment accuracy "Low"  and Key point limit 1000:  0.0247 [m]
Case 4. Alignment accuracy "Low"  and Key point limit 3125:  0.4056 [m]

These are just typical examples selected from 300+ combinations of settings I have tested. The large errors in Cases 1, 2 and 4 are due to large overestimations of ground altitude, associated with the underestimations of the intrinsic parameter f. The result indicates that using large-scale & "selected" features sometimes improves the accuracy of SfM.
« Last Edit: November 13, 2019, 10:18:55 AM by Ryuseiken »

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14813
    • View Profile
Re: The criterion used in Key Point Limit
« Reply #1 on: November 12, 2019, 06:02:29 PM »
Hello Ryuseiken,

Would it be possible to share the source image set and the project with the mentioned cases for our internal tests? If so, please send the download link to support@agisoft.com.

Also I could suggest to check, if you are observing the similar difference if "Fit additional corrections" is used in the Optimize Cameras dialog in the version 1.6.0 pre-release for the same cases.
Best regards,
Alexey Pasumansky,
Agisoft LLC

Ryuseiken

  • Newbie
  • *
  • Posts: 21
    • View Profile
Re: The criterion used in Key Point Limit
« Reply #2 on: November 13, 2019, 05:18:02 AM »
Hello Alexey,

Thank you for your reply.
I will check the 1.6.0 new feature.

Yes, it is possible if the upload size problem (the 664 images have the file size of 5.5 GB) is solved and if you give some feedback.
And if possible, could you tell what kind of keypoints are  excluded in "Key Point Limit" feature? Smaller and less distinct ones?


Ryuseiken

  • Newbie
  • *
  • Posts: 21
    • View Profile
Re: The criterion used in Key Point Limit
« Reply #3 on: November 13, 2019, 10:36:48 AM »
I am sorry but in the four cases I mentioned in my first post, the Cases 1, 2, 4 and the Case 3 had different camera group settings.
When I used the completely same settings except the key point limit and alignment accuracy, the result was:

Case 1. Alignment accuracy "High" and Key point limit 50000: 0.1069 [m]
Case 3. Alignment accuracy "Low"  and Key point limit 1000:  0.0247 [m]
Case 4. Alignment accuracy "Low"  and Key point limit 3125:  0.0380 [m]

I apologize for the misleading post.
Anyway, the key point limit has a significant effect on SfM accuracy (check point errors).

In the attached Excel file I share the results of all 819 runs so far.
Runs 799, 303, and 783 correspond to the three cases listed above.

In addition, I have uploaded the reports for the three cases here:
https://drive.google.com/file/d/1D9p5wqhFJJxE_WDU1K_g_AvJgA_mp4nL/view?usp=sharing.

The version I used is Metashape Professional Version 1.5.5 build 9097 (64 bit).
Thank you for your attention.

« Last Edit: November 13, 2019, 10:56:13 AM by Ryuseiken »

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14813
    • View Profile
Re: The criterion used in Key Point Limit
« Reply #4 on: November 13, 2019, 03:26:14 PM »
Hello Ryuseiken,

Using considerably lower key point / tie point limits is likely acting as significant increase of the camera location coordinates weight for the optimization procedure.
Best regards,
Alexey Pasumansky,
Agisoft LLC

Ryuseiken

  • Newbie
  • *
  • Posts: 21
    • View Profile
Re: The criterion used in Key Point Limit
« Reply #5 on: November 14, 2019, 03:07:41 PM »
Hello Alexey,

Thank you for your reply.
Then, we will check if other measures to increase the relative weight of measured camera coordinates (i.e., decreasing camera accuracy [m] increasing tie point accuracy [pix] settings) in BA have the similar effect to enhance SfM accuracy.