Upsampling a sparse pointcloud

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Upsampling a sparse pointcloud

M4rio
Dear community,

i want to do upsampling on my sparse cloud.
[my goal is to compute descriptors on it, which are then matched against descriptors of a kinect scan of the car in a real scene and do object recognition like proposed in the global pipeline.]

As the kinect scans aren't as sparse as my model cloud, i want to reach a higher resolution. in the picture you can see, the points on the long side of the car seem to lie on curves/lines. i want to get rid of this unnatural structure and have a more dense cloud as a result of upsampling.

do you have any suggestions?
help is really appreciated!


Kind Regards,

Mario

----------------------------------------------------------------
Bachelor student, OTH Regensburg (University of Applied Sciences)
Faculty of Computer Science and Mathematics
Reply | Threaded
Open this post in threaded view
|

Re: Upsampling a sparse pointcloud

VictorLamoine
Administrator
Hello,

You are most probably exporting this point cloud from a CAD software.
If so I would recommend exporting a mesh and using the pcl_mesh_sampling program (https://github.com/PointCloudLibrary/pcl/blob/master/tools/mesh_sampling.cpp) that allows to export a dense, uniform point cloud from a mesh.

Bye