There is a class of algorithms for triangulation between two planar contours. These algorithms try to make a "good triangulation" to fill a space between these contours:
One of them (Optimal surface reconstruction from planar contours) is based on the dynamic programming technique and uses a cost function for determining which triangulation is acceptable according to the minimum cost.
A minimal triangle area as a cost function makes a good result in most of cases (Triangulation of Branching Contours using Area Minimization), but, unfortunately, not in all of them.
For example when you have two rectangle contours that are shifted from each other.
As you can see, according to the minimal area criteria, all points from the contour \alpha will be connected to the point A of the contour \beta, that is not true (a correct triangulation must be a "tube" through both curves, instead of two tetrahedrons).
So my questions are:
1) Does any algorithm that deals with two contours better than the dynamic programming based one exist?
2) If not, which criteria for the cost function can provide a better result?