Recover the reprojection error from Levenberg Marquardt iteration OpenCV

Question

I'm working currently on Camera Calibration using OpenCV C++ library which seems to be very good. As it is mentionned in openCV documentation , the calibrateCamera runs LM optimization to minimize the reprojection error by refining the homography. I understand the theory, however, I must recover the data calculated at each LM iteration to plot the evolution of reprojection error at each iteration, and I really cannot see how to do that, here is a figure that shows an example, I am aiming at the residuals just like in figure a (found only by searching on google).

Answer 1

If you want information out an OpenCV API that it's not designed to give (here, reprojection error for every iteration), you will have to change OpenCV code and recompile the library.