bool make_statistic::generic_run()
{
typedef viennagrid::segmented_mesh<MeshT, SegmentationT> SegmentedMeshType;
typename viennamesh::result_of::const_parameter_handle<SegmentedMeshType>::type imp = input_mesh.get<SegmentedMeshType>();
if (imp)
{
typedef viennamesh::statistic<double> StatisticType;
typedef typename viennagrid::result_of::cell<MeshT>::type CellType;
output_parameter_proxy<StatisticType> osp(output_statistic);
if (histogram_bins.valid())
{
osp().set_histogram( StatisticType::histogram_type::make(histogram_bins().begin(), histogram_bins().end()) );
}
else if (histogram_min.valid() && histogram_max.valid() && histogram_bin_count.valid())
{
osp().set_histogram( StatisticType::histogram_type::make_uniform(histogram_min(), histogram_max(), histogram_bin_count()) );
}
else
{
error(1) << "No histogram configuration provided" << std::endl;
return false;
}
try
{
if (metric_type() == "aspect_ratio")
osp()( imp().mesh, viennamesh::aspect_ratio<CellType> );
else if (metric_type() == "condition_number")
osp()( imp().mesh, viennamesh::condition_number<CellType> );
else if (metric_type() == "min_angle")
osp()( imp().mesh, viennamesh::min_angle<CellType> );
else if (metric_type() == "max_angle")
osp()( imp().mesh, viennamesh::max_angle<CellType> );
else if (metric_type() == "min_dihedral_angle")
osp()( imp().mesh, viennamesh::min_dihedral_angle<CellType> );
else if (metric_type() == "radius_edge_ratio")
osp()( imp().mesh, viennamesh::radius_edge_ratio<CellType> );
else
{
error(1) << "Metric type \"" << metric_type() << "\" is not supported" << std::endl;
return false;
}
}
catch ( metric_not_implemented_or_supported_exception const & ex )
{
error(1) << "Metric type \"" << metric_type() << "\" is not supported" << std::endl;
error(1) << ex.what() << std::endl;
}
info(5) << osp() << std::endl;
return true;
}
return false;
}
bool make_statistic::run(viennamesh::algorithm_handle &)
{
info(1) << "make_statistic started:" << std::endl;
mesh_handle input_mesh = get_required_input<mesh_handle>("mesh");
/*original mesh for comparison statistics (hausdorff distance, curvature difference...)*/
mesh_handle original_mesh = get_input<mesh_handle>("original_mesh");
/*All provided cell quality metric types are implemented as header files, which are located in statistics/metrics.
* Note that most metrics are only implemented for triangles*/
data_handle<viennamesh_string> metric_type = get_required_input<viennamesh_string>("metric_type");
/*Decision threshold for triangle qualitity classification
* E.g., radius_ratio < 1.5
* Whether '<' or '>' is used for comparison is deduced from metric_ordering_tag*/
data_handle<viennagrid_numeric> good_element_threshold = get_input<viennagrid_numeric>("good_element_threshold");
/* comprehensive mesh quality metric is defined as
alpha * (1 - good_elements_counted/count) + beta * min_dist_rms + gamma * mean_curvature + delta * volume_deviation;
*/
data_handle<viennagrid_numeric> alpha = get_input<viennagrid_numeric>("alpha");
data_handle<viennagrid_numeric> beta = get_input<viennagrid_numeric>("beta");
data_handle<viennagrid_numeric> gamma = get_input<viennagrid_numeric>("gamma");
data_handle<viennagrid_numeric> delta = get_input<viennagrid_numeric>("delta");
#if HIST
data_handle<viennagrid_numeric> histogram_bins = get_input<viennagrid_numeric>("histogram_bin");
data_handle<viennagrid_numeric> histogram_min = get_input<viennagrid_numeric>("histogram_min");
data_handle<viennagrid_numeric> histogram_max = get_input<viennagrid_numeric>("histogram_max");
data_handle<int> histogram_bin_count = get_input<int>("histogram_bin_count");
#endif
typedef viennagrid::mesh MeshType;
typedef viennagrid::result_of::element<MeshType>::type ElementType; //=Triangle or Tetrahedron
typedef viennamesh::statistic<viennagrid_numeric> StatisticType;
StatisticType statistic;
#if HIST
typedef StatisticType::histogram_type HistogramType;
if (histogram_bins.valid())
{
std::vector<viennagrid_numeric> bins;
for (int i = 0; i != histogram_bins.size(); ++i)
bins.push_back( histogram_bins(i) );
statistic.set_histogram( StatisticType::histogram_type::make(bins.begin(), bins.end()) );
}
else if (histogram_min.valid() && histogram_max.valid() && histogram_bin_count.valid())
{
statistic.set_histogram( StatisticType::histogram_type::make_uniform(histogram_min(), histogram_max(), histogram_bin_count()) );
}
else
{
//If histograms are about to be added again, returing false here will prevent the basic statistics features from working!
error(1) << "No histogram configuration provided" << std::endl;
return false;
}
#endif
//Good element threshold input was set, call statistics with the appropriate metric
if(good_element_threshold.valid())
{
viennamesh::LoggingStack stack( std::string("High quality cell counter with metric type \"") + metric_type() + "\"" );
if (metric_type() == "aspect_ratio")
statistic.cell_quality_count<viennamesh::aspect_ratio_tag>( input_mesh(), viennamesh::aspect_ratio<ElementType>, good_element_threshold());
else if (metric_type() == "min_angle")
statistic.cell_quality_count<viennamesh::min_angle_tag>( input_mesh(), viennamesh::min_angle<ElementType>, good_element_threshold());
else if (metric_type() == "max_angle")
statistic.cell_quality_count<viennamesh::max_angle_tag>( input_mesh(), viennamesh::max_angle<ElementType>, good_element_threshold());
else if (metric_type() == "min_dihedral_angle")
statistic.cell_quality_count<viennamesh::min_dihedral_angle_tag>( input_mesh(), viennamesh::min_dihedral_angle<ElementType>, good_element_threshold());
else if (metric_type() == "radius_edge_ratio")
statistic.cell_quality_count<viennamesh::radius_edge_ratio_tag>( input_mesh(), viennamesh::radius_edge_ratio<ElementType>, good_element_threshold());
else if (metric_type() == "radius_ratio")
statistic.cell_quality_count<viennamesh::radius_ratio_tag>( input_mesh(), viennamesh::radius_ratio<ElementType>, good_element_threshold());
else if (metric_type() == "perimeter_inradius_ratio")
statistic.cell_quality_count<viennamesh::perimeter_inradius_ratio_tag>( input_mesh(), viennamesh::perimeter_inradius_ratio<ElementType>, good_element_threshold());
else if (metric_type() == "edge_ratio")
statistic.cell_quality_count<viennamesh::edge_ratio_tag>( input_mesh(), viennamesh::edge_ratio<ElementType>, good_element_threshold());
else if (metric_type() == "circum_perimeter_ratio")
statistic.cell_quality_count<viennamesh::circum_perimeter_ratio_tag>( input_mesh(), viennamesh::circum_perimeter_ratio<ElementType>, good_element_threshold());
else if (metric_type() == "stretch")
statistic.cell_quality_count<viennamesh::stretch_tag>( input_mesh(), viennamesh::stretch<ElementType>, good_element_threshold());
else if (metric_type() == "skewness")
statistic.cell_quality_count<viennamesh::skewness_tag>( input_mesh(), viennamesh::skewness<ElementType>, good_element_threshold());
else
{
error(1) << "Metric type \"" << metric_type() << "\" is not supported for cell quality classifaction" << std::endl;
return false;
}
}
else //no triangle classifiction takes place
{
viennamesh::LoggingStack stack( std::string("Cell statistics with metric type \"") + metric_type() + "\"" );
if (metric_type() == "aspect_ratio")
statistic.cell_stats( input_mesh(), viennamesh::aspect_ratio<ElementType> );
else if (metric_type() == "min_angle")
statistic.cell_stats( input_mesh(), viennamesh::min_angle<ElementType> );
else if (metric_type() == "max_angle")
statistic.cell_stats( input_mesh(), viennamesh::max_angle<ElementType> );
else if (metric_type() == "min_dihedral_angle")
statistic.cell_stats( input_mesh(), viennamesh::min_dihedral_angle<ElementType> );
else if (metric_type() == "radius_edge_ratio")
statistic.cell_stats( input_mesh(), viennamesh::radius_edge_ratio<ElementType> );
else if (metric_type() == "radius_ratio")
statistic.cell_stats( input_mesh(), viennamesh::radius_ratio<ElementType> );
else if (metric_type() == "perimeter_inradius_ratio")
statistic.cell_stats( input_mesh(), viennamesh::perimeter_inradius_ratio<ElementType> );
else if (metric_type() == "edge_ratio")
statistic.cell_stats( input_mesh(), viennamesh::edge_ratio<ElementType> );
else if (metric_type() == "circum_perimeter_ratio")
statistic.cell_stats( input_mesh(), viennamesh::circum_perimeter_ratio<ElementType> );
else if (metric_type() == "stretch")
statistic.cell_stats( input_mesh(), viennamesh::stretch<ElementType> );
else if (metric_type() == "skewness")
statistic.cell_stats( input_mesh(), viennamesh::skewness<ElementType> );
else
{
error(1) << "Metric type \"" << metric_type() << "\" is not supported" << std::endl;
return false;
}
}
if(original_mesh.valid())//a second mesh is set, calculate mesh comparison measures
{
viennamesh::LoggingStack stack( std::string("Calculation of Mesh Comparison Measures") );
statistic.mesh_comparison_quality(input_mesh(), original_mesh());
ConstTriangleRange tr_orig(original_mesh());
ConstTriangleRange tr(input_mesh());
StatisticType statistic_orig;
statistic_orig.cell_stats( original_mesh(), viennamesh::aspect_ratio<ElementType> );
//use median of orig mesh for input mesh triangle shape characterization
statistic.cell_quality_count<viennamesh::aspect_ratio_tag>( input_mesh(), viennamesh::aspect_ratio<ElementType>, statistic_orig.median());
if(alpha.valid() && beta.valid() && gamma.valid() && delta.valid() )
{
statistic.set_mesh_quality_weights(alpha(), beta(), gamma(), delta());
info(5) << "values for comprehensive mesh quality metric: alpha = " << alpha()
<< ", beta = " << beta() << ", gamma = " << gamma() << ", delta = "<< delta() << std::endl;
}
else
{
info(5) << "default values for comprehensive mesh quality metric used: alpha = 0.25, beta = 20, gamma = 1.0, delta = 1.3" << std::endl;
}
set_output("minimum_distance_rms", statistic.min_dist_rms());
set_output("mean_curvature_difference", statistic.mean_curvature());
set_output("area_deviation", statistic.volume_deviation());
set_output("triangle_shape", statistic.good_elements()/statistic.count());
set_output("mesh_quality_metric", statistic.mesh_quality_metric());
}
info(5) << statistic << "\n";
#if HIST
statistic.normalize();
std::vector<viennagrid_numeric> bins;
for (HistogramType::const_iterator bit = statistic.histogram().begin(); bit != statistic.histogram().end(); ++bit)
bins.push_back( (*bit).second );
bins.push_back( statistic.histogram().overflow_bin() );
data_handle<viennagrid_numeric> output_bins = make_data<viennagrid_numeric>();
output_bins.set( bins );
set_output( "bins", output_bins );
#endif
set_output( "min", statistic.min() );
set_output( "max", statistic.max() );
set_output( "mean", statistic.mean() );
set_output( "median", statistic.median());
return true;
}
