Frequency domain representations of surfaces

We present "surface frequency" representations, which describe how Cartesian positions on a surface vary with the intrinsic position along the surface. These representations are constructed using optimized coordinate values, individual sample weights, and an iterative transform algorithm. We then use surface frequency to monitor a deformable surface model and increase the number of samples when there is sufficient evidence of undersampling. The resulting model starts with fewer samples, but achieves comparable target fidelity.