3D coordinates are currently calculated assuming that the depth information given in the distance image is the distance between the object point and the corresponding pixel on the chip.
According to PMD informations, however, this depth information represents the distance between the object point and the pinhole of the camera system (origin of ToF world coordinate system).
The according method in ToFProcessingCommon has to be adapted.
| Status | Assigned | Task | ||
|---|---|---|---|---|
| Resolved | None | T12699 Extend mitkToFDistanceImageToPointSetFilterTest and mitkToFDistanceImageToSurfaceFilterTest | ||
| Restricted Maniphest Task | ||||
| Resolved | None | T9302 Correct coordinate calculation in ToFProcessingCommon |
[7e39c7]: Merge branch 'bug-9302-CorrectToFCoordinateCalculation'
2011-09-07 19:01:44 Alexander Seitel [429ebe]
Adapt ToFProcessingCommonTest to changes in coordinate calculation
2011-09-07 19:01:04 Alexander Seitel [0e3e4d]
Changed calculation of 3D coordinates now assuming that the distance given by the image is from the pinhole to the object pont
2011-09-07 19:00:23 Alexander Seitel [72a860]
Adapted documentation graphic to new coordinate calculation
When assuming that the distance image contains the distance between the object point and the camera pinhole, the interpixel distance is not necessary anymore. The calculation "on the side of the camera" (meaning between pinhole and chip) can be done in pixel units only.When assuming that the distance image contains the distance between the object point and the camera pinhole, the interpixel distance is not necessary anymore. The calculation "on the side of the camera" (meaning between pinhole and chip) can be done in pixel units only.
[5302f7]: Merge branch 'bug-9302-correct_coordinate_calculation_in_ToFProcessing
2012-07-21 15:38:56 Anja Groch [fb6bae]
Adapted the tests. In mitkToFProcessingCommonTest both calculation (with and without interpixdist) are tested and also that they produce the same result. The mitkToFDistanceImageToPointSetFilterTest and mitkToFDistanceImageToSurfaceFilterTest still need to cover the case without interpixel distance. This could be done during Bugsquashing.
2012-07-21 15:35:25 Anja Groch [8ddc35]
Changed calculation from distance to cartesian coordinates and vice versa in such way, that the interpixel is not needed anymore. Instead, the focal length needs to be read in pixel units. The old version is still available, if you set m_ReconstructionMode to false.
2012-07-16 17:06:33 Anja Groch [d385f2]
safety commit
[91ac72]: Merge branch 'bug-9302-correct_coordinate_calculation_in_ToFProcessing
2012-07-22 00:21:29 Anja Groch [6811bc]
COMP: forgot to set reconstructionMode to false in the two tests. Also fixed two variable definitions which cause a crash.
[846ad4]: Merge branch 'bug-9302-correct_coordinate_calculation_fix_error_in_doc
2012-07-27 13:55:54 Anja Groch [7b7e79]
Fixed errors in documentation.
[4c3598]: Merge branch 'bug-9302-correct_coordinate_calculation_index_error_fix'
2012-08-23 10:36:07 Anja Groch [19d1bc]
Added GetReconstructionMode() method
2012-08-23 10:34:39 Anja Groch [0120b2]
calculation error due to wrong indices in inline-function fixed