Outlier Robust ICP for Minimizing Fractional RMSD

摘要

We describe a variation of the iterative closest point (ICP) algorithm foraligning two point sets under a set of transformations. Our algorithm issuperior to previous algorithms because (1) in determining the optimalalignment, it identifies and discards likely outliers in a statistically robustmanner, and (2) it is guaranteed to converge to a locally optimal solution. Tothis end, we formalize a new distance measure, fractional root mean squareddistance (frmsd), which incorporates the fraction of inliers into the distancefunction. We lay out a specific implementation, but our framework can easilyincorporate most techniques and heuristics from modern registration algorithms.We experimentally validate our algorithm against previous techniques on 2 and 3dimensional data exposed to a variety of outlier types.