Hi, I'm trying to understand the energy minimization code but having a hard time doing so. From the paper, Gauss-Newton is used for the optimization problem. While reading the code, it seems that it is used at multiple places, epipolar search of immature points, estimation of rigid body motion and brightness transfer parameters in coarse tracking of new frame, and in joint optimization of active window. I'm not well versed in optimization so I'm having a lot of trouble understanding this last part. What are the roles of AccumulatedSCHessian, AccumulatedTopHessian etc. ?
Is there anyone who could provide a high level (beginner friendly) explanation of how the energy functional code works?
Hi, I'm trying to understand the energy minimization code but having a hard time doing so. From the paper, Gauss-Newton is used for the optimization problem. While reading the code, it seems that it is used at multiple places, epipolar search of immature points, estimation of rigid body motion and brightness transfer parameters in coarse tracking of new frame, and in joint optimization of active window. I'm not well versed in optimization so I'm having a lot of trouble understanding this last part. What are the roles of
AccumulatedSCHessian
,AccumulatedTopHessian
etc. ?Is there anyone who could provide a high level (beginner friendly) explanation of how the energy functional code works?