Skip to main content

GeometryInfo

Namespace: DISP3DLIB  ·  Library: 3D Display Library

#include <disp3D/geometryinfo.h>

class DISP3DLIB::GeometryInfo

This class allows sensor-to-mesh mapping and calculation of surface constrained distances.

This class holds static methods for sensor-to-mesh mapping and surface constrained distance calculation on a mesh

src/examples/ex_disp3d_scene/main.cpp
MatrixX3f sensors(2, 3);
sensors << -0.2f, 0.1f, 0.5f,                                                            // above vertex 0
    3.1f, 2.9f, 0.4f;                                                                    // above vertex 15
VectorXi projected = GeometryInfo::projectSensors(vertices, sensors);                    // nearest vertex per sensor
QSharedPointer<MatrixXd> distances = GeometryInfo::scdc(vertices, neighbors, projected); // geodesic: n_vertices x n_sensors

Public Methods​

GeometryInfo()​

Static Methods​

scdc(matVertices, vecNeighborVertices, vecVertSubset, dCancelDist)​

scdc Calculates surface constrained distances on a mesh.

Parameters:

  • matVertices : const Eigen::MatrixX3f & Vertex positions (nVertices x 3).

  • vecNeighborVertices : const std::vector< Eigen::VectorXi > & Adjacency list for each vertex.

  • vecVertSubset : Eigen::VectorXi & Source vertex indices; if empty it is filled with all vertex indices.

  • dCancelDist : double Maximum geodesic distance to explore, in vertex coordinate units.

Returns:

  • QSharedPointer<Eigen::MatrixXd> — Distance table (nVertices x nSubset); entries beyond dCancelDist stay infinite.

scdcInterpolationMat(matVertices, vecNeighborVertices, vecVertSubset, interpolationFunction, dCancelDist, progressCallback, cancelledFlag)​

scdcInterpolationMat Computes geodesic distances (SCDC) and builds the sparse interpolation matrix in a single pass, without allocating a dense nVertices x nSources distance table.

Produces identical results to scdc() + Interpolation::createInterpolationMat() but uses dramatically less memory (sparse representation only stores entries within cancelDist).

Parameters:

  • matVertices : const Eigen::MatrixX3f & Vertex positions (nVertices x 3).

  • vecNeighborVertices : const std::vector< Eigen::VectorXi > & Adjacency list for each vertex.

  • vecVertSubset : const Eigen::VectorXi & Source vertex indices.

  • interpolationFunction : double()(double)* Weight function (e.g. Interpolation::cubic).

  • dCancelDist : double Maximum geodesic distance for interpolation.

  • progressCallback : std::function< void(int, int)> Optional callback reporting Dijkstra progress (current, total).

  • cancelledFlag : *const std::atomic< bool > ** Optional pointer to an atomic bool; when set to true the computation aborts early.

Returns:

  • QSharedPointer< Eigen::SparseMatrix< float > > — Sparse interpolation matrix (nVertices x nSources), or empty matrix if cancelled.

projectSensors(matVertices, matSensorPositions)​

projectSensors Calculates the nearest neighbor vertex to each sensor.

Parameters:

  • matVertices : const Eigen::MatrixX3f & Vertex positions (nVertices x 3).

  • matSensorPositions : const Eigen::MatrixX3f & Sensor positions (nSensors x 3) in the same coordinate frame.

Returns:

  • Eigen::VectorXi — Index of the nearest mesh vertex for each sensor.

filterBadChannels(matDistanceTable, fiffInfo, iSensorType)​

filterBadChannels Filters bad channels from distance table.

Parameters:

  • matDistanceTable : QSharedPointer<Eigen::MatrixXd> Distance table whose bad-channel columns are set to infinity.

  • fiffInfo : const FiffInfo & Measurement info providing channel list and bad channel names.

  • iSensorType : qint32 FIFF channel kind selecting the sensors that form the table columns.

Returns:

  • Eigen::VectorXi — Column indices of the bad channels that were found.

Example​

Used in examples​

ex_disp3d_scene​

Disp3D scene data: sensor-to-surface interpolation, electrodes, MRI slices, video overlay and layers.

src/examples/ex_disp3d_scene/main.cpp, line 115
MatrixX3f sensors(2, 3);
sensors << -0.2f, 0.1f, 0.5f,                                                            // above vertex 0
    3.1f, 2.9f, 0.4f;                                                                    // above vertex 15
VectorXi projected = GeometryInfo::projectSensors(vertices, sensors);                    // nearest vertex per sensor
QSharedPointer<MatrixXd> distances = GeometryInfo::scdc(vertices, neighbors, projected); // geodesic: n_vertices x n_sensors
// scipy.sparse.csgraph.dijkstra on the same graph: Manhattan distances to vertices 0 and 15
ok &= expect(projected == Vector2i(0, 15) && distances->rows() == 16 && (*distances)(5, 0) == 2.0 && (*distances)(5, 1) == 4.0 && (*distances)(3, 0) == 3.0 && (*distances)(3, 1) == 3.0,
             "GeometryInfo projects the sensors onto vertices 0 and 15; geodesic distances match scipy");

ex_interpolation​

Example of using the interpolation library and geometryInfo library.

src/examples/ex_interpolation/main.cpp, line 130
//projecting with MEG
qint64 startTimeProjecting = QDateTime::currentMSecsSinceEpoch();
VectorXi mappedSubSet = GeometryInfo::projectSensors(t_sensorSurfaceVV[0].rr, megSensors);
std::cout << "Projecting duration: " << QDateTime::currentMSecsSinceEpoch() - startTimeProjecting << " ms " << std::endl;

Authors of this file​