xpppq / python-pcl

Python bindings to the pointcloud library (pcl)

⚠⚠ A new maintainer for this library is sought. Please see strawlab#395. ⚠⚠

Introduction

This is a small python binding to the pointcloud library. Currently, the following parts of the API are wrapped (all methods operate on PointXYZ) point types

• I/O and integration; saving and loading PCD files
• segmentation
• SAC
• smoothing
• filtering
• registration (ICP, GICP, ICP_NL)

The code tries to follow the Point Cloud API, and also provides helper function for interacting with NumPy. For example (from tests/test.py)

import pcl
import numpy as np
p = pcl.PointCloud(np.array([[1, 2, 3], [3, 4, 5]], dtype=np.float32))
seg = p.make_segmenter()
seg.set_model_type(pcl.SACMODEL_PLANE)
seg.set_method_type(pcl.SAC_RANSAC)
indices, model = seg.segment()

or, for smoothing

import pcl
p = pcl.load("C/table_scene_lms400.pcd")
fil = p.make_statistical_outlier_filter()
fil.set_mean_k (50)
fil.set_std_dev_mul_thresh (1.0)
fil.filter().to_file("inliers.pcd")

Point clouds can be viewed as NumPy arrays, so modifying them is possible using all the familiar NumPy functionality:

import numpy as np
import pcl
p = pcl.PointCloud(10)  # "empty" point cloud
a = np.asarray(p)       # NumPy view on the cloud
a[:] = 0                # fill with zeros
print(p[3])             # prints (0.0, 0.0, 0.0)
a[:, 0] = 1             # set x coordinates to 1
print(p[3])             # prints (1.0, 0.0, 0.0)

More samples can be found in the examples directory, and in the unit tests.

This work was supported by Strawlab.

Requirements

This release has been tested on Linux Ubuntu 16.04 with

• Python 2.7.6, 3.5.x
• pcl 1.7.2(apt install)
• Cython <= 0.25.2

This release has been tested on Linux Ubuntu 18.04 with

• Python 2.7.6, 3.5.x
• pcl 1.8.1(apt install)
• Cython <= 0.25.2

and MacOS with

• Python 2.7.6, 3.5.x
• pcl 1.9.1(use homebrew)
• Cython <= 0.25.2

and Windows with

• (Miniconda/Anaconda) - Python 3.4
• pcl 1.6.0(VS2010)
• Cython <= 0.25.2
• Gtk+

and Windows with

• (Miniconda/Anaconda) - Python 3.5
• pcl 1.8.1(VS2015)
• Cython <= 0.25.2
• Gtk+

and Windows with

• (Miniconda/Anaconda) - Python 3.6
• pcl 1.8.1(VS2017[Priority High]/VS2015[not VS2017 Install])
• Cython == 0.25.2
• Gtk+

Installation

Linux(Ubuntu)

before Install module

Ubuntu16.04/18.04 (use official package)

1. Install PCL Module.

$ sudo apt-get update -y

$ sudo apt-get install libpcl-dev -y

Reference `here <https://packages.ubuntu.com/search?keywords=libpcl-dev>`_.

PCL 1.8.x/1.9.x and Ubuntu16.04/18.04(build module)([CI Test Timeout])

1. Build Module

   Reference here.

MacOSX

before Install module

Case1. use homebrew(PCL 1.9.1 - 2018/12/25 current)

1. Install PCL Module.

   $ brew tap homebrew/science

   $ brew install pcl

Case1. use old homebrew(PCL 1.8.1 - 2017/11/13 current)

1. Check git log.

   $ cd /usr/local/Library/Formula

   $ git log ...

2. git checkout (target hash) pcl.rb

   write after.

Warning:

Current Installer (2017/10/02) Not generated pcl-2d-1.8.pc file.(Issue #119)

Reference PointCloudLibrary Issue.

Pull request 1679.

Issue 1978.

circumvent:

copy travis/pcl-2d-1.8.pc file to /usr/local/lib/pkgconfig folder.

Windows

Using pip with a precompiled wheel

This is the simpliest method on windows. The wheel contains the PCL binaries _ and thus you do not need to install the original PCL library.

1. Go in the history on the appveyor page
2. Click on the last successful revision (green) and click on the job corresponding to your python version
3. Go in the artfacts section for that job and download the wheel (the file with extension whl)
4. In the command line, move to your download folder and run the following command (replacing XXX by the right string)

pip install python_pcl-XXX.whl

Compiling the binding from source

If the method using the procompiled wheel does not work you can compile the binding from the source.

before Install module

Case1. use PCL 1.6.0

Windows SDK 7.1

            PCL All-In-One Installer

            32 bit

64 bit

           OpenNI2[(PCL Install FolderPath)\3rdParty\OpenNI\OpenNI-(win32/x64)-1.3.2-Dev.msi]

Case2. use 1.8.1/1.9.1

            Visual Studio 2015 C++ Compiler Tools(use Python 2.7/3.5/3.6/3.7)

            Visual Studio 2017 C++ Compiler Tools(use Python 3.6.x/3.7.x)

            PCL All-In-One Installer

1.8.1

            Visual Studio 2015 - 32 bit

            Visual Studio 2017 - 32 bit

            Visual Studio 2015 - 64 bit

            Visual Studio 2017 - 64 bit

1.9.1

            Visual Studio 2017 - 32 bit

            Visual Studio 2017 - 64 bit

          OpenNI2[(PCL Install FolderPath)\3rdParty\OpenNI2\OpenNI-Windows-(win32/x64)-2.2.msi]

        Common setting

Windows Gtk+ Download             Download file unzip. Copy bin Folder to pkg-config Folder

Download file unzip. Copy bin Folder to pkg-config Folder

or execute powershell file [Install-GTKPlus.ps1].

Python Version use VisualStudio Compiler

set before Environment variable

1. PCL_ROOT

set PCL_ROOT=(PCL Install/Build_Binary FolderPath)

2. PATH

(pcl 1.6.0)
set PATH=%PCL_ROOT%/bin/;%OPEN_NI_ROOT%/Tools;%VTK_ROOT%/bin;%PATH%

(pcl 1.8.1/1.9.1)
set PATH=%PCL_ROOT%/bin/;%OPEN_NI2_ROOT%/Tools;%VTK_ROOT%/bin;%PATH%

Common setting

1. pip module install.

pip install --upgrade pip

pip install cython

pip install numpy

2. install python module

   python setup.py build_ext -i

   python setup.py install

3. install python-pcl with conda (solved)

-> conda create -n ipk # create a new conda env. -> conda activate ipk # activate env.

-> conda update -n base -c defaults conda # update conda

-> conda config --add channels conda-forge # add conda-forge channels -> conda install -c sirokujira python-pcl # pcl installation -> conda install -c jithinpr2 gtk3 # Gtk+ Gui dependency -> conda install -y ipython # install ipython -> conda install -y jupyter # install jupyter

After that, run jupyter notebook or ipython shell to test pcl installation.

Build & Test Status

windows(1.6.0/1.8.1/1.9.1)

  .. image:: https://ci.appveyor.com/api/projects/status/w52fee7j22q211cm/branch/master?svg=true

target

https://ci.appveyor.com/project/Sirokujira/python-pcl-iju42

Mac OSX(1.9.1)/Ubuntu16.04(1.7.2)

A note about types

Point Cloud is a heavily templated API, and consequently mapping this into Python using Cython is challenging.

It is written in Cython, and implements enough hard bits of the API (from Cythons perspective, i.e the template/smart_ptr bits) to provide a foundation for someone wishing to carry on.

API Documentation

Read the docs.

For deficiencies in this documentation, please consult the PCL API docs, and the PCL tutorials.