Polynomial reconstruction of 3D sampled curves using auxiliary surface data

This paper proposes a method for structural enhancement of a 3D sampled curve. The curve is assumed to be organized, but corrupted with low frequency noise. The proposed method approaches the notion of curve reconstruction in a novel way, where information about the structure in a scanned surface is used to reconstruct the curve. Principal Component Analysis is carried out on successive neighborhoods along the curve to estimate reduced dimensionality spaces, which allows polynomial reconstruction. The effectiveness of the proposed method is verified by both simulations and experiments.