Skip to content

Latest commit

 

History

History
64 lines (42 loc) · 2.78 KB

install-pcl.md

File metadata and controls

64 lines (42 loc) · 2.78 KB
Raw

Install PCL

We use the Point Coud Library (PCL) for 2D/3D image and point cloud processing. Official download page: link

Install PCL 1.7.2-1+trusty6 with VTK 5.8.0-14.1ubuntu3 (Ubuntu 14.04)

For Trusty and similar versions, we need to add the jochen-sprickerhof PPA [ref], while VTK is available directly as part of universe [ref].

sudo add-apt-repository ppa:v-launchpad-jochen-sprickerhof-de/pcl
sudo apt update
sudo apt install libpcl-all libpcl-all-dev
sudo apt install libopenni-sensor-primesense0 # no pcl-tools here; and libopenni-sensor-pointclouds0 cannot be installed simultaneously.

Install PCL 1.7.2-3~utopic1 with VTK (Ubuntu 14.10)

For Utopic and similar versions, we need to add the jochen-sprickerhof PPA [ref].

sudo add-apt-repository ppa:v-launchpad-jochen-sprickerhof-de/pcl
sudo apt update
sudo apt install libpcl-dev
sudo apt install pcl-tools # no sensor here, but libopenni-sensor-primesense0 may be found elsewhere

Install PCL 1.7.2-14build1 with VTK 6.2.0+dfsg1-10build1 (Ubuntu 16.04)

PCL is available directly as part of universe in modern Ubuntu distros [ref].

sudo apt install libpcl-dev  # depends: libvtk6-dev

Developers may find obscure No rule to make target '/usr/lib/x86_64-linux-gnu/libproj.so' errors in their PCL-dependent applications, this is due to a known bug. Workaround (SO answer):

sudo apt install libproj-dev

Besides, remove a superfluous vtkproj4 item from the list of PCL libraries in your CMake code:

find_package(PCL 1.7) # you probably have this around somewhere

# add these lines afterwards
if(DEFINED PCL_LIBRARIES)
    list(REMOVE_ITEM PCL_LIBRARIES "vtkproj4")
endif()

See also PointCloudLibrary/pcl#2406 (linker error in case new point types are added).

Install PCL (From source)

Link: install from source.

Install PCL (For any distro with CUDA for KinFu)

No official ppa, install CUDA, then install from source setting additional flags to get pcl_kinfu_largeScale, pcl_kinfu_largeScale_mesh_output, pcl_kinfu_largeScale_texture_output.

See additional info here.