Object Matching Algorithm Based on Robust Hausdorff Distance

周志强,汪渤

DOI: https://doi.org/10.3724/sp.j.1087.2009.00086

2009-01-01

Journal of Computer Applications

Abstract:Based on the conventional Hausdorff distance of edge position, we introduced the information of edge gradient to distance measuring, and constructed a new distance function that used the information of edge position and gradient. Then we presented a robust three-dimension Hausdorff distance and the object matching algorithm using the proposed Hausdorff distance. In the proposed algorithm, we used a two-step matching scheme that combined two steps of coarse and refined matching to solve the problem of computation complexity increasing caused by using 3D distance measure. Experimental result shows that the matching robustness of the proposed algorithm has been improved greatly compared to conventional algorithms that only use the information of edge position.