pompeiu_hausdorff

Code for "Cascading upper bounds for triangle soup Pompeiu-Hausdorff distance" (SGP 2024).
Authors: Leonardo Sacht and Alec Jacobson.

C++

-------- Compilation --------

mkdir build
cd build
cmake ..
make

-------- Example usage --------

./pompeiu_hausdorff ../meshes/107100.obj ../meshes/107100_sf.obj 1e-8 1000000 1

-------- Input ----------
argv[1]: path to triangle soup A in .obj format
argv[2]: path to triangle soup B in .obj format
argv[3]: tolerance for the difference between upper and lower bound
argv[4]: factor to define the maximum allowed number of faces and vertices in the subdivided mesh A with respect to the number of faces and vertices of the initial mesh A
argv[5]: 0 (false) or 1 (true) to normalize tolerance by the length of the diagonal of A's bounding box
-------- Output (printed) ----------
lower bound (absolute and relative to dA), upper bound (absolute and relative to dA), and timings

Python

On python you can install with

pip install cascading_upper_bounds

Then use, for example, like this:

from cascading_upper_bounds import pompeiu_hausdorff
import numpy as np
import igl

VA, FA = igl.read_triangle_mesh(f"meshes/107100.obj")
VB, FB = igl.read_triangle_mesh(f"meshes/107100_sf.obj")
tol = 1e-8
max_factor = 1000000.0
normalize = True
lower, upper_max, dA, time_taken_bvh, time_taken_bounds = pompeiu_hausdorff(VA, FA, VB, FB, tol, max_factor, normalize)

print("dA: ", dA)
print("lower bound: ", lower)
print("upper bound: ", upper_max)
print("lower bound/dA: ", lower/dA)
print("upper bound/dA: ", upper_max/dA)
print("time taken bvh(ms): ", time_taken_bvh)
print("time taken bounds(ms): ", time_taken_bounds)