// Poisson reconstruction method:
// Reconstructs a surface mesh from a point set and returns it as a polyhedron.
Polyhedron* poisson_reconstruct(const Point_set& points,
Kernel::FT sm_angle, // Min triangle angle (degrees).
Kernel::FT sm_radius, // Max triangle size w.r.t. point set average spacing.
Kernel::FT sm_distance, // Approximation error w.r.t. point set average spacing.
const QString& solver_name,
bool use_two_passes) // solver name
{
CGAL::Timer task_timer; task_timer.start();
//***************************************
// Checks requirements
//***************************************
if (points.size() == 0)
{
std::cerr << "Error: empty point set" << std::endl;
return NULL;
}
bool points_have_normals = (points.begin()->normal() != CGAL::NULL_VECTOR);
if ( ! points_have_normals )
{
std::cerr << "Input point set not supported: this reconstruction method requires oriented normals" << std::endl;
return NULL;
}
CGAL::Timer reconstruction_timer; reconstruction_timer.start();
//***************************************
// Computes implicit function
//***************************************
std::cerr << "Computes Poisson implicit function "
<< "using " << solver_name.toLatin1().data() << " solver...\n";
// Creates implicit function from the point set.
// Note: this method requires an iterator over points
// + property maps to access each point's position and normal.
// The position property map can be omitted here as we use iterators over Point_3 elements.
Poisson_reconstruction_function function(
points.begin(), points.end(),
CGAL::make_normal_of_point_with_normal_pmap(Point_set::value_type()));
bool ok = false;
#ifdef CGAL_EIGEN3_ENABLED
if(solver_name=="Eigen - built-in simplicial LDLt")
{
CGAL::Eigen_solver_traits<Eigen::SimplicialCholesky<CGAL::Eigen_sparse_matrix<double>::EigenType> > solver;
ok = function.compute_implicit_function(solver, use_two_passes);
}
if(solver_name=="Eigen - built-in CG")
{
CGAL::Eigen_solver_traits<Eigen::ConjugateGradient<CGAL::Eigen_sparse_matrix<double>::EigenType> > solver;
solver.solver().setTolerance(1e-6);
solver.solver().setMaxIterations(1000);
ok = function.compute_implicit_function(solver, use_two_passes);
}
#endif
// Computes the Poisson indicator function f()
// at each vertex of the triangulation.
if ( ! ok )
{
std::cerr << "Error: cannot compute implicit function" << std::endl;
return NULL;
}
// Prints status
std::cerr << "Total implicit function (triangulation+refinement+solver): " << task_timer.time() << " seconds\n";
task_timer.reset();
//***************************************
// Surface mesh generation
//***************************************
std::cerr << "Surface meshing...\n";
// Computes average spacing
Kernel::FT average_spacing = CGAL::compute_average_spacing(points.begin(), points.end(),
6 /* knn = 1 ring */);
// Gets one point inside the implicit surface
Kernel::Point_3 inner_point = function.get_inner_point();
Kernel::FT inner_point_value = function(inner_point);
if(inner_point_value >= 0.0)
{
std::cerr << "Error: unable to seed (" << inner_point_value << " at inner_point)" << std::endl;
return NULL;
}
// Gets implicit function's radius
Kernel::Sphere_3 bsphere = function.bounding_sphere();
Kernel::FT radius = std::sqrt(bsphere.squared_radius());
// Defines the implicit surface: requires defining a
// conservative bounding sphere centered at inner point.
Kernel::FT sm_sphere_radius = 5.0 * radius;
Kernel::FT sm_dichotomy_error = sm_distance*average_spacing/1000.0; // Dichotomy error must be << sm_distance
Surface_3 surface(function,
Kernel::Sphere_3(inner_point,sm_sphere_radius*sm_sphere_radius),
sm_dichotomy_error/sm_sphere_radius);
// Defines surface mesh generation criteria
CGAL::Surface_mesh_default_criteria_3<STr> criteria(sm_angle, // Min triangle angle (degrees)
sm_radius*average_spacing, // Max triangle size
sm_distance*average_spacing); // Approximation error
CGAL_TRACE_STREAM << " make_surface_mesh(sphere center=("<<inner_point << "),\n"
<< " sphere radius="<<sm_sphere_radius<<",\n"
<< " angle="<<sm_angle << " degrees,\n"
<< " triangle size="<<sm_radius<<" * average spacing="<<sm_radius*average_spacing<<",\n"
<< " distance="<<sm_distance<<" * average spacing="<<sm_distance*average_spacing<<",\n"
<< " dichotomy error=distance/"<<sm_distance*average_spacing/sm_dichotomy_error<<",\n"
<< " Manifold_with_boundary_tag)\n";
// Generates surface mesh with manifold option
STr tr; // 3D Delaunay triangulation for surface mesh generation
C2t3 c2t3(tr); // 2D complex in 3D Delaunay triangulation
CGAL::make_surface_mesh(c2t3, // reconstructed mesh
surface, // implicit surface
criteria, // meshing criteria
CGAL::Manifold_with_boundary_tag()); // require manifold mesh
// Prints status
std::cerr << "Surface meshing: " << task_timer.time() << " seconds, "
<< tr.number_of_vertices() << " output vertices"
<< std::endl;
task_timer.reset();
if(tr.number_of_vertices() == 0)
return NULL;
// Converts to polyhedron
Polyhedron* output_mesh = new Polyhedron;
CGAL::output_surface_facets_to_polyhedron(c2t3, *output_mesh);
// Prints total reconstruction duration
std::cerr << "Total reconstruction (implicit function + meshing): " << reconstruction_timer.time() << " seconds\n";
//***************************************
// Computes reconstruction error
//***************************************
// Constructs AABB tree and computes internal KD-tree
// data structure to accelerate distance queries
AABB_tree tree(faces(*output_mesh).first, faces(*output_mesh).second, *output_mesh);
tree.accelerate_distance_queries();
// Computes distance from each input point to reconstructed mesh
double max_distance = DBL_MIN;
double avg_distance = 0;
for (Point_set::const_iterator p=points.begin(); p!=points.end(); p++)
{
double distance = std::sqrt(tree.squared_distance(*p));
max_distance = (std::max)(max_distance, distance);
avg_distance += distance;
}
avg_distance /= double(points.size());
std::cerr << "Reconstruction error:\n"
<< " max = " << max_distance << " = " << max_distance/average_spacing << " * average spacing\n"
<< " avg = " << avg_distance << " = " << avg_distance/average_spacing << " * average spacing\n";
return output_mesh;
}
bool poisson_reconstruct(FaceGraph* graph,
Point_set& points,
typename Traits::FT sm_angle, // Min triangle angle (degrees).
typename Traits::FT sm_radius, // Max triangle size w.r.t. point set average spacing.
typename Traits::FT sm_distance, // Approximation error w.r.t. point set average spacing.
const QString& solver_name, // solver name
bool use_two_passes,
bool do_not_fill_holes)
{
// Poisson implicit function
typedef CGAL::Poisson_reconstruction_function<Traits> Poisson_reconstruction_function;
// Surface mesher
typedef CGAL::Surface_mesh_default_triangulation_3 STr;
typedef CGAL::Surface_mesh_complex_2_in_triangulation_3<STr> C2t3;
typedef CGAL::Implicit_surface_3<Traits, Poisson_reconstruction_function> Surface_3;
// AABB tree
typedef CGAL::AABB_face_graph_triangle_primitive<FaceGraph> Primitive;
typedef CGAL::AABB_traits<Traits, Primitive> AABB_traits;
typedef CGAL::AABB_tree<AABB_traits> AABB_tree;
CGAL::Timer task_timer; task_timer.start();
//***************************************
// Checks requirements
//***************************************
if (points.size() == 0)
{
std::cerr << "Error: empty point set" << std::endl;
return false;
}
bool points_have_normals = points.has_normal_map();
if ( ! points_have_normals )
{
std::cerr << "Input point set not supported: this reconstruction method requires oriented normals" << std::endl;
return false;
}
CGAL::Timer reconstruction_timer; reconstruction_timer.start();
//***************************************
// Computes implicit function
//***************************************
std::cerr << "Computes Poisson implicit function "
<< "using " << solver_name.toLatin1().data() << " solver...\n";
// Creates implicit function from the point set.
// Note: this method requires an iterator over points
// + property maps to access each point's position and normal.
Poisson_reconstruction_function function(points.begin_or_selection_begin(), points.end(),
points.point_map(), points.normal_map());
bool ok = false;
#ifdef CGAL_EIGEN3_ENABLED
if(solver_name=="Eigen - built-in simplicial LDLt")
{
CGAL::Eigen_solver_traits<Eigen::SimplicialCholesky<CGAL::Eigen_sparse_matrix<double>::EigenType> > solver;
ok = function.compute_implicit_function(solver, use_two_passes);
}
if(solver_name=="Eigen - built-in CG")
{
CGAL::Eigen_solver_traits<Eigen::ConjugateGradient<CGAL::Eigen_sparse_matrix<double>::EigenType> > solver;
solver.solver().setTolerance(1e-6);
solver.solver().setMaxIterations(1000);
ok = function.compute_implicit_function(solver, use_two_passes);
}
#endif
// Computes the Poisson indicator function f()
// at each vertex of the triangulation.
if ( ! ok )
{
std::cerr << "Error: cannot compute implicit function" << std::endl;
return false;
}
// Prints status
std::cerr << "Total implicit function (triangulation+refinement+solver): " << task_timer.time() << " seconds\n";
task_timer.reset();
//***************************************
// Surface mesh generation
//***************************************
std::cerr << "Surface meshing...\n";
// Computes average spacing
Kernel::FT average_spacing = CGAL::compute_average_spacing<Concurrency_tag>(points.all_or_selection_if_not_empty(),
6 /* knn = 1 ring */,
points.parameters());
// Gets one point inside the implicit surface
Kernel::Point_3 inner_point = function.get_inner_point();
Kernel::FT inner_point_value = function(inner_point);
if(inner_point_value >= 0.0)
{
std::cerr << "Error: unable to seed (" << inner_point_value << " at inner_point)" << std::endl;
return false;
}
// Gets implicit function's radius
Kernel::Sphere_3 bsphere = function.bounding_sphere();
Kernel::FT radius = std::sqrt(bsphere.squared_radius());
// Defines the implicit surface: requires defining a
// conservative bounding sphere centered at inner point.
Kernel::FT sm_sphere_radius = 5.0 * radius;
Kernel::FT sm_dichotomy_error = sm_distance*average_spacing/1000.0; // Dichotomy error must be << sm_distance
Surface_3 surface(function,
Kernel::Sphere_3(inner_point,sm_sphere_radius*sm_sphere_radius),
sm_dichotomy_error/sm_sphere_radius);
// Defines surface mesh generation criteria
CGAL::Surface_mesh_default_criteria_3<STr> criteria(sm_angle, // Min triangle angle (degrees)
sm_radius*average_spacing, // Max triangle size
sm_distance*average_spacing); // Approximation error
CGAL_TRACE_STREAM << " make_surface_mesh(sphere center=("<<inner_point << "),\n"
<< " sphere radius="<<sm_sphere_radius<<",\n"
<< " angle="<<sm_angle << " degrees,\n"
<< " triangle size="<<sm_radius<<" * average spacing="<<sm_radius*average_spacing<<",\n"
<< " distance="<<sm_distance<<" * average spacing="<<sm_distance*average_spacing<<",\n"
<< " dichotomy error=distance/"<<sm_distance*average_spacing/sm_dichotomy_error<<",\n"
<< " Manifold_with_boundary_tag)\n";
// Generates surface mesh with manifold option
STr tr; // 3D Delaunay triangulation for surface mesh generation
C2t3 c2t3(tr); // 2D complex in 3D Delaunay triangulation
CGAL::make_surface_mesh(c2t3, // reconstructed mesh
surface, // implicit surface
criteria, // meshing criteria
CGAL::Manifold_with_boundary_tag()); // require manifold mesh
// Prints status
std::cerr << "Surface meshing: " << task_timer.time() << " seconds, "
<< tr.number_of_vertices() << " output vertices"
<< std::endl;
task_timer.reset();
if(tr.number_of_vertices() == 0)
return false;
// Converts to polyhedron
CGAL::facets_in_complex_2_to_triangle_mesh(c2t3, *graph);
// Prints total reconstruction duration
std::cerr << "Total reconstruction (implicit function + meshing): " << reconstruction_timer.time() << " seconds\n";
//***************************************
// Computes reconstruction error
//***************************************
// Constructs AABB tree and computes internal KD-tree
// data structure to accelerate distance queries
AABB_tree tree(faces(*graph).first, faces(*graph).second, *graph);
tree.accelerate_distance_queries();
// Computes distance from each input point to reconstructed mesh
double max_distance = DBL_MIN;
double avg_distance = 0;
std::set<typename boost::graph_traits<FaceGraph>::face_descriptor> faces_to_keep;
for (Point_set::const_iterator p=points.begin_or_selection_begin(); p!=points.end(); p++)
{
typename AABB_traits::Point_and_primitive_id pap = tree.closest_point_and_primitive (points.point (*p));
double distance = std::sqrt(CGAL::squared_distance (pap.first, points.point(*p)));
max_distance = (std::max)(max_distance, distance);
avg_distance += distance;
typename boost::graph_traits<FaceGraph>::face_descriptor f = pap.second;
faces_to_keep.insert (f);
}
avg_distance /= double(points.size());
std::cerr << "Reconstruction error:\n"
<< " max = " << max_distance << " = " << max_distance/average_spacing << " * average spacing\n"
<< " avg = " << avg_distance << " = " << avg_distance/average_spacing << " * average spacing\n";
if (do_not_fill_holes)
{
typename boost::graph_traits<FaceGraph>::face_iterator it = faces(*graph).begin ();
while (it != faces(*graph).end ())
{
typename boost::graph_traits<FaceGraph>::face_iterator current = it ++;
if (faces_to_keep.find (*current) == faces_to_keep.end ())
{
CGAL::Euler::remove_face(halfedge (*current, *graph), *graph);
}
}
}
return true;
}
