Hello,
I'm not sure I understand how we handle the case of multiple disparity maps (more than 2 photos). In slide Stereo, 28-29 it seems that we merge the disparities picture and then apply the variational approach. Am I correct? However, in the lecture "Shape from Stereo-2", minute 29:00, the professor says that every disparity map gives a cloud of 3D points and that, in the end, we will fit a surface on the set of points given by all disparity maps at hand. Do we do such fit using the variational approach ( Stereo, slide 30 ) to allow for discontinuities? This would mean that we don't average among all disparity maps at hand, but we directly use the variational method ???
