Generating 3D point clouds from RGB+D observations (CObservation3DRangeScan objects)

Contents

1. Calibration parameters
2. Projection equations
- 2.1. General case
- 2.2. “Rectified” depth image
3. Related MRPT APIs

1. Calibration parameters

Firstly, you must have very clear the calibration parameters involved in the process.

Each “RGB+D observation” is stored in MRPT as one object of type mrpt::slam::CObservation3DRangeScan (click to see its full description), which for this tutorial will be assumed that contains:

An RGB image (in: mrpt::img::CImage intensityImage;) and its camera parameters (in: mrpt::img::TCamera cameraParams; ).
A depth image. In: mrpt::math::CMatrix rangeImage;) and its camera parameters (in: mrpt::img::TCamera cameraParamsIntensity;).
The relative 6D pose between the two cameras (in: mrpt::poses::CPose3D relativePoseIntensityWRTDepth;).

These observations can be captured from a sensor (e.g. Kinect) in real-time, or loaded from a dataset, as explained here.

In any case, the calibration parameters of both cameras will come already filled in with their correct values (or, at least, those the user provided in the moment of capturing/grabbing the dataset!).

This set of three pieces of data that must be calibrated (for example, see the tutorial for Kinect calibration) before generating precise 3D point clouds from RGB+D observations are: the two sets of camera parametersand the relative 6D pose between them.

3D camera reference systems

2. Projection equations

(Write me!) In the meanwhile, please refer to the source code of CObservation3DRangeScan_project3D_impl.h and to the example kinect_online_offline_demo.

2.1. General case

…

2.2. “Rectified” depth image

…

3. Related MRPT APIs

There exist different possibilities:

3.1. RGB+D –> local 3D point cloud

CObservation3DRangeScanPtr obsRGBD = ...  // Get a smart pointer to the RGB+D observation data from wherever 
obsRGBD->project3DPointsFromDepthImage(); // Project points into the internal buffers of "obsRGBD"

// Points are now in the object buffers as std::vector<float>: 
cout << obsRGBD->points3D_x[0] << endl;
cout << obsRGBD->points3D_y[0] << endl;
cout << obsRGBD->points3D_z[0] << endl;

CObservation3DRangeScanPtr obsRGBD = ... // Get a smart pointer to the RGB+D observation data from wherever

obsRGBD->project3DPointsFromDepthImage(); // Project points into the internal buffers of "obsRGBD"

// Points are now in the object buffers as std::vector<float>:

cout << obsRGBD->points3D_x[0] << endl;

cout << obsRGBD->points3D_y[0] << endl;

cout << obsRGBD->points3D_z[0] << endl;

3.2. RGB+D –> local 3D point cloud –> CPointsMap (& derived)

CObservation3DRangeScanPtr obsRGBD = ...  // Get a smart pointer to the RGB+D observation data from wherever 
obsRGBD->project3DPointsFromDepthImage(); // Project points into the internal buffers of "obsRGBD"

// Load into an XYZ point map: 
CSimplePointsMap pntsMap;
pntsMap.minDistBetweenLaserPoints = 0.05; // In meters
pntsMap.loadFromRangeScan(*obsRGBD);

// *OR* into an XYZ+RGB point map:
CColouredPointsMap pntsMap;
pntsMap.minDistBetweenLaserPoints = 0.05; // In meters
pntsMap.colorScheme.scheme = CColouredPointsMap::cmFromIntensityImage;
pntsMap.loadFromRangeScan(*obsRGBD);

CObservation3DRangeScanPtr obsRGBD = ... // Get a smart pointer to the RGB+D observation data from wherever

obsRGBD->project3DPointsFromDepthImage(); // Project points into the internal buffers of "obsRGBD"

// Load into an XYZ point map:

CSimplePointsMap pntsMap;

pntsMap.minDistBetweenLaserPoints = 0.05; // In meters

pntsMap.loadFromRangeScan(*obsRGBD);

// *OR* into an XYZ+RGB point map:

CColouredPointsMap pntsMap;

pntsMap.minDistBetweenLaserPoints = 0.05; // In meters

pntsMap.colorScheme.scheme = CColouredPointsMap::cmFromIntensityImage;

pntsMap.loadFromRangeScan(*obsRGBD);

3.3. RGB+D –> CPointsMap (& derived)

CObservation3DRangeScanPtr obsRGBD = ...  // Get a smart pointer to the RGB+D observation data from wherever 
CSimplePointsMap pntsMap;  // Or any other class derived from CPointsMap. If it has RGB data it will get filled.
obsRGBD->project3DPointsFromDepthImageInto( 
  pntsMap, 
  false /* without obs.sensorPose */ 
  );

CObservation3DRangeScanPtr obsRGBD = ... // Get a smart pointer to the RGB+D observation data from wherever

CSimplePointsMap pntsMap; // Or any other class derived from CPointsMap. If it has RGB data it will get filled.

obsRGBD->project3DPointsFromDepthImageInto(

pntsMap,

false /* without obs.sensorPose */

);

3.4. RGB+D –> mrpt::opengl::CPointCloud

CObservation3DRangeScanPtr obsRGBD = ...  // Get a smart pointer to the RGB+D observation data from wherever
mrpt::opengl::CPointCloudPtr gl_points = mrpt::opengl::CPointCloud::Create();  // Smart pointer to opengl point cloud (with XYZ data)
obsRGBD->project3DPointsFromDepthImageInto(
&nbsp;&nbsp;&nbsp; *gl_points,
&nbsp;&nbsp;&nbsp; false /* without obs.sensorPose */
&nbsp;&nbsp;&nbsp; );

CObservation3DRangeScanPtr obsRGBD = ... // Get a smart pointer to the RGB+D observation data from wherever

mrpt::opengl::CPointCloudPtr gl_points = mrpt::opengl::CPointCloud::Create(); // Smart pointer to opengl point cloud (with XYZ data)

obsRGBD->project3DPointsFromDepthImageInto(

    *gl_points,

    false /* without obs.sensorPose */

    );

3.5. RGB+D –> mrpt::opengl::CPointCloudColoured

CObservation3DRangeScanPtr obsRGBD = ...&nbsp; // Get a smart pointer to the RGB+D observation data from wherever
mrpt::opengl::CPointCloudColouredPtr gl_points = mrpt::opengl::CPointCloudColoured::Create(); // Smart pointer to opengl point cloud (with XYZ & RGB data)
obsRGBD->project3DPointsFromDepthImageInto(
   *gl_points,
   false /* without obs.sensorPose */
   );

CObservation3DRangeScanPtr obsRGBD = ...  // Get a smart pointer to the RGB+D observation data from wherever

mrpt::opengl::CPointCloudColouredPtr gl_points = mrpt::opengl::CPointCloudColoured::Create(); // Smart pointer to opengl point cloud (with XYZ & RGB data)

obsRGBD->project3DPointsFromDepthImageInto(

*gl_points,

false /* without obs.sensorPose */

);

3.6. RGB+D –> pcl::PointCloud<PointXYZ>

CObservation3DRangeScanPtr obsRGBD = ...  // Get a smart pointer to the RGB+D observation data from wherever 
pcl::PointCloud&lt;pcl::PointXYZ&gt; cloud;
obsRGBD-&gt;project3DPointsFromDepthImageInto(
    cloud,
    false /* without obs.sensorPose */
    );

CObservation3DRangeScanPtr obsRGBD = ... // Get a smart pointer to the RGB+D observation data from wherever

pcl::PointCloud<pcl::PointXYZ> cloud;

obsRGBD->project3DPointsFromDepthImageInto(

cloud,

false /* without obs.sensorPose */

);

3.7. RGB+D –> pcl::PointCloud<PointXYZRGB>

CObservation3DRangeScanPtr obsRGBD = ...&nbsp; // Get a smart pointer to the RGB+D observation data from wherever
pcl::PointCloud<pcl::PointXYZRGB> cloud;
obsRGBD->project3DPointsFromDepthImageInto(
  cloud,
  false /* without obs.sensorPose */
  );

CObservation3DRangeScanPtr obsRGBD = ...  // Get a smart pointer to the RGB+D observation data from wherever

pcl::PointCloud<pcl::PointXYZRGB> cloud;

obsRGBD->project3DPointsFromDepthImageInto(

cloud,

false /* without obs.sensorPose */

);