In the description of the optimization problem in the wiki, the non-negativity constraint is not
explicitly mentioned. I think most people will apply it automatically, but strictly speaking it
is in general not clear that the solution with the best cost is non-negative everywhere.
On the other hand, I think the RDP prior only makes sense for non-negative images, but
this is not obvious.
In the description of the optimization problem in the wiki, the non-negativity constraint is not explicitly mentioned. I think most people will apply it automatically, but strictly speaking it is in general not clear that the solution with the best cost is non-negative everywhere.
On the other hand, I think the RDP prior only makes sense for non-negative images, but this is not obvious.