Fractured Surfaces Matching for Reassembling Broken Solids

A fractured surfaces matching algorithm for reassembling broken solids is proposed. the algorithm first defines point bumpiness based on integral invariants that is robust to noise. Using the bumpiness feature points are picked from fractured surface. Subsequently initial matching point pairs are obtained by comparing multi-scale integral invariants of feature points and outliers are discarded according to surface patches similarity based on compatibility constraint. Finally, a fast and efficient hierarchical greedy algorithm is employed to align the two fractured surfaces and measure if they are matched using surface distance. the algorithm is also suited for matching arbitrary surfaces which contain rich feature points.