4

I'm looking into the reports & the Mechanism of UAV Photogrammetry software like Pix4D and AgiSoft.

These two words are mentioned in a lot of support documentation, but they are defined no-where.

From my studies, I remember that tie-points are points marking the same physical feature in two or more images.

What does key-point mean? Is it the same?

Improve this question

2 Answers 2

Reset to default
3

I found this page in the Pix4D documentation which states:

Keypoints are points of interest (high contrast, interesting texture) on the images that can be easily recognized. The number of keypoints depends on:

  • The size of the images.
  • The visual content.

A 14MP image will generate between 5'000 and 50'000 keypoints per image.

And this page which defines manual and automatic tie points.

An Automatic Tie Points is a 3D point and its corresponding 2D keypoints that were automatically detected in the images and used to compute its 3D position.

A Manual Tie Point is a point without 3D coordinates that is marked by the user in the images. It can be used to asses and improve the reconstruction accuracy.

Improve this answer
0

Key-Points are what is being recognized between the images. In Pix4D for example if there are dominant features in the images, it would be easier for software to recognize and tie them together, which would generally result higher key-points.

Therefore for snow, high-densed vegetation and water, you usually end up with lower key-points because it is difficult for software to tie the images with each other.

Moreover in Pix4D processing options like image scale and calibration method also affects the number of key-points. You can also force Pix4D to limit the number of key-points.

Improve this answer
2
  • What do you mean when you say: 'Key-Points are what is being recognized between the images' ? Isn't that what tie points are?
    – Devdatta Tengshe
    Commented Apr 25, 2018 at 6:07
  • 1
    Hey, sorry for not being clear enough. What I meant was key points are based on dominant, and recognizable features from images. As an example if there are cars or buildings in the images , it would be easier for software to recognize and to stitch it compare to snow high-dense vegetation or water, because all the images would look similar. Tie-points also contribute to key-points in Pix4D, like if you add manual-tie points it would most likely increase the number of key-points as well. Moreover key-points are usually determined by the sensor and the resolution.
    – Selim Erguden
    Commented Apr 25, 2018 at 11:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.