I think you have the right idea.
The top and bottom surface need 'matching' parameterizations -- whatever that means.
Airfoil files are usually specified with Y values at certain X points. The obvious parameterization for that is an X- parameterization. When you take an airfoil file and average matching top / bottom points, you get a certain mean camber line.
I parameterize the upper and lower curves with an arc-length parameterization (scaled to [0,1] for each). Then, equal parameter points are paired and averaged. This mean camber line will be slightly different than the equal x-coordinate.
Neither one is the true mean camber line (as defined by thin airfoil theory) but that takes more work to calculate -- and I haven't gone to the trouble yet. Someday...
Rob