Camera calibration while tracking a natural marker

Question

Do I need to calibrate the camera if I have to implement a natural marker tracker?

In fact, I do not quite understand the idea of ​​calibrating the camera, although I read that it is necessary to increase 3d / 2d objects in the image feed.

Answer 1

Calibrating the camera means searching for the internal parameters of the camera. It is necessary, of course, if you want to successfully find natural functions here , I will explain why.

Then you need to look only for external parameters . You only need to perform the calibration once, since the camera is always the same (given that you cannot zoom in / out, change the focal length, etc.). Without calibrating the camera, you will have many problems for the task of tracking natural functions, since it is more complicated than tracking fiduciary remedies.

In the link I gave you, you will also find how to calculate a pose from a planar marker. This is theoretical, but you can find a lot of code on the Internet. If you need more help, tell me, I can explain in more detail if necessary.

Answer 2

Strictly speaking, you can detect functions, match patterns to recognize a marker, and then track these functions without calibrating the camera. Calibration allows us to determine both internal (for example, distortion coefficients) and external camera parameters (for example, rotation), which are required when someone needs to determine the boundaries of markers or evaluate 3D poses.

Answer 3

It's necessary? No.

This is useful? You are betting. A rule of thumb is ALWAYS if you can calibrate the camera for your stationary camera, do it.

You can do many things with such information: remove distortion, get distance in some metric space ... Most trackers have a basic assumption / models, these models are best suited when the data is in a space where the model makes sense. Camera calibration is an easy way to achieve this.