Robust Hausdorff distance

The function modifies the Hausdorff distance to deal with noise and outliers

The original Hausdorff distance (HD) is highly sensitive to outliers and noise. In particular, under these undesirable conditions, HD usually overestimates the distance between pairs of point sets. In this submission, a technique is proposed to overcome the challenge. We have proposed kernels that remove erroneous data while promoting higher accuracy of the HD. To run the codes, you can easily run the files demoMain.m, contained in the parent directory, and demoCompare.m, contained in the directory ../Comparisons (MHD Vs RHD).
A manuscript for this submission has been submitted to the Journal of Pattern Recognition for consideration.