Blended Intrinsic Maps

The paper proposes a fully automatic approach for finding a mapping between two nonisometric surfaces. The approach is to blend low-dimensional conformal maps with weights varying smoothly over the surface. The advantage of the approach is that it can handle a wide variety of deformations while keeping the search space of the possible maps polynomial. The paper first details how to find candidate conformal maps. Second, it defines the smooth blending weights that reflect the area preservation of a map at every point p. Then, a global optimisation problem of finding which candidate maps to blend is solved using their area preservation and consistency measures. Experimental results show that the proposed approach produces better surface matchings than alternative approaches over variety of data sets.