SyB3R - Synthetic Benchmark for 3D Reconstruction
|
Classes | |
struct | Statistics |
bundles several statistical values (minimal, maximal, average values as well as standard deviation, and number of all and of correctly estimated points More... | |
Public Member Functions | |
PointCloudEvaluation (const std::vector< Eigen::Vector3f > &refPcld, const std::vector< bool > &objectOfInterest, const std::vector< Eigen::Vector3f > &estPcld, const Eigen::Matrix4f &estToRefTrafo, float thresh) | |
computation of performance measures More... | |
const Statistics & | getPCCorrectness () const |
const Statistics & | getPCCompleteness () const |
const std::vector< float > & | getPerPointCorrectness () const |
returns correctness per point, i.e. distance to closest point in reference point cloud | |
const std::vector< float > & | getPerPointCompleteness () const |
returns completeness per point, i.e. distance to closest point in estimated point cloud | |
const Eigen::Matrix4f & | getRefinedEstToRefTrafo () const |
returns used transformation matrix from coordinate system of the estimate to the reference | |
float | get95PercentPrecision () const |
returns precision value p such that 95% of the points have a precision larger than p | |
float | get95PercentCompleteness () const |
returns completeness value c such that 95% of the points have a completeness larger than c | |
void | createMatlabPlots (std::ostream &stream, float precCutoff, float complCutoff, float metricScale) const |
creates a script to generate matlab plots summarizing the computed statistics | |
Protected Member Functions | |
void | nnAnalysis (pcl::octree::OctreePointCloudSearch< pcl::PointXYZ > &, pcl::PointCloud< pcl::PointXYZ >::Ptr, float, Statistics &, const std::vector< bool > &objectOfInterest, std::vector< float > &distances, bool computeCompleteness) |
nearest neighbor analyis between two point clouds | |
void | computeCorrectness (pcl::octree::OctreePointCloudSearch< pcl::PointXYZ > &, pcl::PointCloud< pcl::PointXYZ >::Ptr, float, const std::vector< bool > &objectOfInterest) |
distance from each point in estimated pc to closest point in reference pc | |
void | computeCompleteness (pcl::octree::OctreePointCloudSearch< pcl::PointXYZ > &, pcl::PointCloud< pcl::PointXYZ >::Ptr, float, const std::vector< bool > &objectOfInterest) |
distance from each point in reference pc to closest point in estimated pc | |
void | eigen2pcl (const std::vector< Eigen::Vector3f > &, pcl::PointCloud< pcl::PointXYZ >::Ptr) |
void | transformPointCloud (pcl::PointCloud< pcl::PointXYZ >::Ptr, const Eigen::Matrix4f &) |
Static Protected Member Functions | |
static void | transformPointCloud (std::vector< Eigen::Vector3f > &points, const Eigen::Matrix4f &transform) |
Protected Attributes | |
Statistics | m_cloudCorrectness |
statistics over correctness | |
Statistics | m_cloudCompleteness |
statistics over completeness | |
std::vector< float > | m_correctness |
std::vector< float > | m_completeness |
Eigen::Matrix4f | m_estToRefTransform |
4x4 transformation matrix from coordinate system of the estimate to the reference | |
struct syb3r::analysis::PointCloudEvaluation::Statistics |
syb3r::analysis::PointCloudEvaluation::PointCloudEvaluation | ( | const std::vector< Eigen::Vector3f > & | refPcld, |
const std::vector< bool > & | objectOfInterest, | ||
const std::vector< Eigen::Vector3f > & | estPcld, | ||
const Eigen::Matrix4f & | estToRefTrafo, | ||
float | thresh | ||
) |
computation of performance measures
refPcld | Reference point cloud |
objectOfInterest | denotes which pixels belong to the object of interest. |
estPcld | Estimated point cloud |
estToRefTrafo | 4x4 tranformation matrix from coordinate system of the estimate to the reference |
thresh | Maximal distance between points in both point clouds to count them as "correct" match |