computer vision - How is the reprojection error calculated in Matlab's triangulate function? Sadly, the documentation gives no mathematical formula -

How is the error of reproducing a triangular function in Matlab?

Unfortunately, there is no mathematical formula.

It only says: There is an average repetition error in the vector for every M world point.

What is the process / uses Matlab while calculating this error?

I have searched the SOF but nothing could be found on this IMHO important question.

UPDATE: How they can use this error to filter bad matches from here:

The AFAIK reprojection error is always calculated in the same way (in the field in general, of computer vision).

There is an error between the interrupter (by name) and the reprojected point in the original point.

So you are triangular with 2 (or more) digits in the camera and get 3D scores in the world order. Due to the errors in the calibration of the camera, the point will not be 100% accurate. What you do, the results take the 3d point ( P ) and with camera calibration parameters, it projects again in the camera, the new Find the original people ( p )

then you get the original point and "reprojected" between Euclidean Calculate the distance.

If you use the matlab If you want to know a bit more about the method to be done, then I will also increase the number of pages in which you will be given the page number:

Multiple view geometry in computer vision By Richard Hartley and Andrew Ziserman (p 312) Cambridge University Press, 2003.

But basically it is at least one of the least intersections, which has no geographical explanation.