Approximating Gromov-Hausdorff distance in Euclidean space

Sushovan Majhi,Jeffrey Vitter,Carola Wenk

DOI: https://doi.org/10.1016/j.comgeo.2023.102034

IF: 0.455

2024-01-01

Computational Geometry

Abstract:The Gromov-Hausdorff distance ( d G H ) proves to be a useful distance measure between shapes. In order to approximate d G H for X , Y ⊂ R d , we look into its relationship with d H , i s o , the infimum Hausdorff distance under Euclidean isometries. As already known for dimension d ≥ 2 , d H , i s o cannot be bounded above by a constant factor times d G H . For d = 1 , however, we prove that d H , i s o ≤ 5 4 d G H . We also show that the bound is tight. In effect, for X , Y ⊂ R with at most n points, this gives rise to an O ( n log ⁡ n ) -time algorithm to approximate d G H ( X , Y ) with an approximation factor of ( 1 + 1 4 ) .

mathematics, applied