Why only left part of velodyne point cloud can be read?

[WeiyiLi]

Since this is a problem I encountered after I run the code, I started a new issue.

When I run the code, I always got message
[pcl::SampleConsensusModel::getSamples] Can not select 0 unique points out of 0!
[pcl::RandomSampleConsensus::computeModel] No samples cound be selected!

from some of ransac.computModel() ,ransac_l.computeModel() and ransac_sphere.computeModel(); functionn in Calibration3DMarker.cpp

I guess it is because I did not have enough data to process, so I tried to display the result of project function in Calibration3DMarker.cpp(line 30), it looks like below

I can see that only the very left is read and processed so none of the spheres can be detected. Even I tried to increase the size of Rect, it still cannot include all the data, especially the four circles in the middle of the data.

How can I process all the data from velodyne but not only the left part?

Thx!

[martin-velas]

Hi! Can you post also the camera image? I would recommend you to move the marker physically to the left :). If it would cause marker to be outside of camera view frustum, you can pre-process the input point cloud by additional rotation around vertical Y axis (and compose this transformation with estimated calibration afterwards).
Hope it helps. Cheers, Martin.