使用可选依赖项配置 C++ 示例

本教程介绍如何使用 CMake 为 Zivid C++ 示例配置可选依赖项。

要求

可选要求

一些示例依赖于外部库，特别是：

备注

以下说明已使用这些版本的依赖项进行了测试：

Visual Studio 2019 及更高版本
CMake 3.24 及更高版本
PCL 1.12.1
Eigen 3.4.0
OpenCV 4.7.0

如果您使用其他版本，您可能会遇到问题。

根据您要运行的示例程序安装所需的依赖项；请参阅下表。

示例程序	PCL	Eigen	OpenCV	OpenCV (CUDA)	Halcon
ReadPCLVis3D	YES
CaptureWritePCLVis3D	YES
CaptureAndVisualizeNormals	YES
StitchByTransformation		YES
StitchByTransformationFromZDF		YES
StitchViaLocalPointCloudRegistration		YES
StitchContinuouslyRotatingObject		YES
StitchUsingRobotMountedCamera		YES
MaskPointCloud			YES
Capture2DAnd3D	YES		YES
ROIBoxViaCheckerboard		YES
UtilizeHandEyeCalibration		YES
PoseConversions		YES
ROIBoxViaArucoMarker		YES
CaptureUndistort2D			YES
CreateDepthMap			YES
ProjectAndFindMarker			YES
ReprojectPoints			YES
ReadProjectAndCaptureImage			YES
GammaCorrection			YES
AllocateMemoryForPointCloudData			YES
CaptureAndConvertImageWithOpenCVOnCuda				YES
TransformPointCloudViaCheckerboard			YES
TransformPointCloudViaArucoMarker			YES
CaptureHalconViaGenICam					YES
CaptureHalconViaZivid					YES

Ubuntu

Eigen

通过运行以下命令从包管理器安装 Eigen 3：

sudo apt install libeigen3-dev

导航到您在 zivid-cpp-samples 中创建的构建目录并运行

mkdir build
cd build
cmake -DEIGEN3_INCLUDE_DIR=/usr/include/eigen3 -DUSE_OPENCV=OFF -DUSE_PCL=OFF ../source
cmake --build .

PCL

通过运行以下命令从包管理器安装 PCL：

sudo apt install libpcl-dev

导航到您在 zivid-cpp-samples 中创建的构建目录并运行

mkdir build
cd build
cmake -DUSE_PCL=ON -DUSE_EIGEN3=OFF -DUSE_OPENCV=OFF ../source
cmake --build .

OpenCV

通过运行以下命令从包管理器安装 OpenCV：

sudo apt install libopencv-dev clang-tidy

导航到您在 zivid-cpp-samples 中创建的构建目录并运行

mkdir build
cd build
cmake -DUSE_OPENCV=ON -DUSE_EIGEN3=OFF -DUSE_PCL=OFF ../source
cmake --build .

OpenCV (CUDA)

To install CUDA follow the instructions on the NVIDIA linux guide.

Older graphics cards may not support all the CUDA instructions required and will require additional configuration. You can find more information on the official OpenCV CUDA introduction and OpenCV contrib repository.

警告

If you encounter random system freezes while using cmake --build ., try adding the --parallel n option with n lower than your CPU core count.

Install tools needed to build OpenCV with CUDA support by running:

sudo apt install clang-tidy ninja-build cmake

Build OpenCV with CUDA support:

wget https://github.com/opencv/opencv/archive/refs/tags/4.13.0.zip
unzip 4.13.0.zip -d opencv
rm 4.13.0.zip
wget https://github.com/opencv/opencv_contrib/archive/refs/tags/4.13.0.zip
unzip 4.13.0.zip -d opencv
rm 4.13.0.zip
cd opencv/opencv-4.13.0
mkdir build
cd build
cmake -S .. -D WITH_CUDA=ON -G Ninja -D CMAKE_BUILD_TYPE=Release -D OPENCV_EXTRA_MODULES_PATH=../../opencv_contrib-4.13.0/modules/ -D ENABLE_CUDA_FIRST_CLASS_LANGUAGE=ON -D CUDAToolkit_ROOT=/usr/local/cuda
cmake --build .
cmake --install . --prefix local/

Navigate to the build directory you created in zivid-cpp-samples and run:

mkdir build
cd build
cmake -S ../source -D CMAKE_PREFIX_PATH=<path-to-opencv-build>/local/lib/cmake/opencv4 -D USE_OPENCV=ON -D USE_CUDA=ON -D USE_OPENCV_CUDA=ON -D EIGEN3_INCLUDE_DIR=/usr/include/eigen3 -D CUDAToolkit_ROOT=/usr/local/cuda -G Ninja -D CMAKE_BUILD_TYPE=Release -D CMAKE_EXPORT_COMPILE_COMMANDS=ON
cmake --build .

When running the CUDA sample, include the built OpenCV libraries in LD_LIBRARY_PATH:

export LD_LIBRARY_PATH=<path-to-opencv-build>/local/lib:$LD_LIBRARY_PATH

Windows

如果要一次性配置多个依赖项，只需在 CMake 中同时添加多个依赖项即可。例如，如果要使用所有需要 Eigen、PCL 和 OpenCV 的示例，可以运行以下命令：

cmake -DEIGEN3_INCLUDE_DIR=<path-to-eigen> -DUSE_PCL=ON -DOpenCV_DIR=<path-to-opencv> ../source

其中 <path-to-eigen> 和 <path-to-opencv> 分别是 Eigen 和 OpenCV 文件夹的路径。