安装OpenCV3.2.0与CUDA10.2

傷城~ 2022-11-29 11:27 378阅读 0赞

# 建议先安装OpenCV3.2.0，再安装CUDA10.2，否则再安装OpenCV3.2.0时，会出现不兼容的情况，解决方案比较麻烦 #

# 如下记录先安装CUDA10.2，否则再安装OpenCV3.2.0时，出现的问题 #

# 安装CUDA10.2，这里就不介绍了，在本文底部会介绍 #

# 安装OpenCV3.2.0 #

事实证明使用conda便捷安装的opencv是[阉割版][Link 1]，**不能实现视频和摄像头的读取功能**，所以需要自己手动编译。

（1）下载opencv，我在window中用迅雷下载，下载后放到Ubuntu中，这样比较快。

进入官网 : [http://opencv.org/releases.html][http_opencv.org_releases.html] , 选择 3.2.0 版本的 source , 下载 opencv-3.2.0.zip (这里的照片是3.1.0 版本)  
![这里写图片描述][SouthEast]

解压后，放在软件安装目录下，我是放在了“/home/zqzy”目录下了

（2）更新，否则可能会面会报错

sudo apt-get update
    sudo apt-get upgrade

（3）安装依赖

sudo apt-get install build-essential 
    sudo apt-get install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev
    sudo apt-get install python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libjasper-dev libdc1394-22-dev
    sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev libv4l-dev
    sudo apt-get install libxvidcore-dev libx264-dev
    sudo apt-get install libgtk-3-dev
    sudo apt-get install libatlas-base-dev gfortran pylint
    sudo apt-get install python2.7-dev python3.5-dev
    sudo apt-get install build-essential  
    sudo apt-get install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev  
    sudo apt-get install python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libjasper-dev libdc1394-22-dev # 处理图像所需的包  
    sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev libv4l-dev liblapacke-dev
    sudo apt-get install libxvidcore-dev libx264-dev # 处理视频所需的包  
    sudo apt-get install libatlas-base-dev gfortran # 优化opencv功能
    sudo apt-get install ffmpeg 
    
    sudo apt-get install libvtk5-dev
     
    或者
     
    sudo apt-get isntall libvtk6-dev

（4）修改 opencv-3.1.0/modules/cudalegacy/src/graphcuts.cpp 文件**（安装opencv3.2.0不用修改）**

sudo gedit opencv-3.1.0/modules/cudalegacy/src/graphcuts.cpp

如果不修改的话，在执行 make -j8 命令编译到 92% 时可能会出现以下错误：

modules/cudalegacy/src/graphcuts.cpp:120:54: error: 
    ‘NppiGraphcutState’ has not been declared
    typedef NppStatus (*init_func_t)(NppiSize oSize, 
    NppiGraphcutState** ppState, Npp8u* pDeviceMem);

由于opecv3.1与cuda8.0不兼容导致的。解决办法：

修改 opencv-3.1.0/modules/cudalegacy/src/graphcuts.cpp 文件![这里写图片描述][SouthEast 1]内容，如图：

（5）编译

cd  /home/zqzy/opencv-3.1.0
    
    mkdir build # 创建编译的文件目录
    
    cd build
    
    cmake -D CMAKE_BUILD_TYPE=Release -D WITH_VTK=ON -D CMAKE_INSTALL_PREFIX=/usr/local ..
    
    sudo make -j8  #编译

执行下面这行时出错：

cmake -D CMAKE_BUILD_TYPE=Release -D WITH_VTK=ON -D CMAKE_INSTALL_PREFIX=/usr/local ..

由于在编译过程中，需要下载ippicv\_linux\_20151201.tgz，这一般需要连接外网，国内一般连接不上，所以下载失败，导致出现这个问题

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3podWlxaXV6aHVveXVlNTgz_size_16_color_FFFFFF_t_70][]

解决方法：

下载ippicv\_linux\_20151201.tgz，参考：[https://blog.csdn.net/zhuiqiuzhuoyue583/article/details/106620707][https_blog.csdn.net_zhuiqiuzhuoyue583_article_details_106620707]

将下载的ippicv\_linux\_20151201.tgz，放到指定的文件目录下：

/home/zqzy/catkin_ws/src/opencv-3.2.0/3rdparty/ippicv/downloads/linux-808b791a6eac9ed78d32a7666804320e

这个路径是根据出错的提示来创建的。

如果没有，就自己创建这个路径，如果该路径下有个ippicv\_linux\_20151201.tgz，将其删除，再把下载好的ippicv\_linux\_20151201.tgz放进去，因为原来那个不完整。

放入ippicv\_linux\_20151201.tgz后，进入build目录，重新编译

cd  /home/zqzy/opencv-3.1.0
    
    cd build
    
    cmake -D CMAKE_BUILD_TYPE=Release -D WITH_VTK=ON -D CMAKE_INSTALL_PREFIX=/usr/local ..
    
    sudo make -j8  #编译

如果出现如下错误：

CMake Error: The following variables are used in this project, but they are set to NOTFOUND.
    Please set them or make sure they are set and tested correctly in the CMake files:
    CUDA_nppi_LIBRARY (ADVANCED)
    linked by target "opencv_cudev" in directory D:/Cproject/opencv/opencv/sources/modules/cudev
    linked by target "opencv_cudev" in directory D:/Cproject/opencv/opencv/sources/modules/cudev
    linked by target "opencv_test_cudev" in directory D:/Cproject/opencv/opencv/sources/modules/cudev/test
    linked by target "opencv_core" in directory D:/Cproject/opencv/opencv/sources/modules/core
    linked by target "opencv_core" in directory D:/Cproject/opencv/opencv/sources/modules/core
    linked by target "opencv_test_core" in directory D:/Cproject/opencv/opencv/sources/modules/core
    linked by target "opencv_perf_core" in directory D:/Cproject/opencv/opencv/sources/modules/core
    linked by target "opencv_test_cudaarithm" in directory D:/Cproject/opencv/opencv/sources/modules/cudaarithm
    linked by target "opencv_cudaarithm" in directory D:/Cproject/opencv/opencv/sources/modules/cudaarithm
    linked by target "opencv_cudaarithm" in directory D:/Cproject/opencv/opencv/sources/modules/cudaarithm
    linked by target "opencv_perf_cudaarithm" in directory D:/Cproject/opencv/opencv/sources/modules/cudaarithm
    linked by target "opencv_flann" in directory D:/Cproject/opencv/opencv/sources/modules/flann
    linked by target "opencv_flann" in directory D:/Cproject/opencv/opencv/sources/modules/flann
    linked by target "opencv_test_flann" in directory D:/Cproject/opencv/opencv/sources/modules/flann
    linked by target "opencv_imgproc" in directory D:/Cproject/opencv/opencv/sources/modules/imgproc
    .................................
    .............

解决方案如下：

1.在opencv3.2.0的cmake中找到FindCUDA.cmake文件

a..找到行

find_cuda_helper_libs(nppi)

改为

find_cuda_helper_libs(nppial)
      find_cuda_helper_libs(nppicc)
      find_cuda_helper_libs(nppicom)
      find_cuda_helper_libs(nppidei)
      find_cuda_helper_libs(nppif)
      find_cuda_helper_libs(nppig)
      find_cuda_helper_libs(nppim)
      find_cuda_helper_libs(nppist)
      find_cuda_helper_libs(nppisu)
      find_cuda_helper_libs(nppitc)

b.找到行

set(CUDA_npp_LIBRARY "${CUDA_nppc_LIBRARY};${CUDA_nppi_LIBRARY};${CUDA_npps_LIBRARY}")

改为

set(CUDA_npp_LIBRARY "${CUDA_nppc_LIBRARY};${CUDA_nppial_LIBRARY};${CUDA_nppicc_LIBRARY};${CUDA_nppicom_LIBRARY};${CUDA_nppidei_LIBRARY};${CUDA_nppif_LIBRARY};${CUDA_nppig_LIBRARY};${CUDA_nppim_LIBRARY};${CUDA_nppist_LIBRARY};${CUDA_nppisu_LIBRARY};${CUDA_nppitc_LIBRARY};${CUDA_npps_LIBRARY}")

c.找到行

unset(CUDA_nppi_LIBRARY CACHE)

改为

unset(CUDA_nppial_LIBRARY CACHE)
    unset(CUDA_nppicc_LIBRARY CACHE)
    unset(CUDA_nppicom_LIBRARY CACHE)
    unset(CUDA_nppidei_LIBRARY CACHE)
    unset(CUDA_nppif_LIBRARY CACHE)
    unset(CUDA_nppig_LIBRARY CACHE)
    unset(CUDA_nppim_LIBRARY CACHE)
    unset(CUDA_nppist_LIBRARY CACHE)
    unset(CUDA_nppisu_LIBRARY CACHE)
    unset(CUDA_nppitc_LIBRARY CACHE)

2.在opencv3.2.0的cmake中找到文件OpenCVDetectCUDA.cmake

修改以下几行

...
      set(__cuda_arch_ptx "")
      if(CUDA_GENERATION STREQUAL "Fermi")
        set(__cuda_arch_bin "2.0")
      elseif(CUDA_GENERATION STREQUAL "Kepler")
        set(__cuda_arch_bin "3.0 3.5 3.7")
      ...

改为

...
      set(__cuda_arch_ptx "")
      if(CUDA_GENERATION STREQUAL "Kepler")
        set(__cuda_arch_bin "3.0 3.5 3.7")
      elseif(CUDA_GENERATION STREQUAL "Maxwell")
        set(__cuda_arch_bin "5.0 5.2")
      ...

3.将头文件cuda\_fp16.h添加至 opencv3.2.0/modules/cudev/include/opencv2/cudev/common.hpp

即在common.hpp中添加

#include <cuda_fp16.h>

重新编译即可

参考：[https://blog.csdn.net/u014613745/article/details/78310916][https_blog.csdn.net_u014613745_article_details_78310916]

如果出现如下错误：

Unsupported gpu architecture 'compute_20'
    .......

使用如下命令进行cmake

cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local -D CUDA_GENERATION=Kepler ..

**关键就是在最后指定使用Kepler**

参考：[https://blog.csdn.net/renhanchi/article/details/80493232][https_blog.csdn.net_renhanchi_article_details_80493232]

(6)编译成功后安装：

sudo make -j8  #编译
    sudo make install #安装

（7）配置OpenCV环境变量  
    配置环境  
   将opencv的库加入到路径，从而让系统可以找到

sudo gedit /etc/ld.so.conf.d/opencv.conf

或者，如果上面这个文件没有，就修改下面这个文件

sudo gedit /etc/ld.so.conf

末尾加入`/usr/local/lib`，保存退出

执行：

sudo ldconfig    #使配置生效

再执行：

sudo gedit /etc/bash.bashrc

末尾加入

PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib/pkgconfig export PKG_CONFIG_PATH
    export PKG_CONFIG_PATH

保存退出

sudo source /etc/bash.bashrc  #使配置生效

该步骤可能会报错找不到命令，原因是source是root命令

错误如下：

slam2@slam2-vm:~/catkin_ws/src$ sudo source /etc/bash.bashrc 
    [sudo] password for slam2: 
    sudo: source: command not found

解决方法：

执行

su  #（进入root权限）

输入密码

如果出现“su: Authentication failure问题”：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3podWlxaXV6aHVveXVlNTgz_size_16_color_FFFFFF_t_70 1][]

参考：[https://blog.csdn.net/zhuiqiuzhuoyue583/article/details/106626618][https_blog.csdn.net_zhuiqiuzhuoyue583_article_details_106626618]

[https://blog.csdn.net/cv\_you/article/details/77341631][https_blog.csdn.net_cv_you_article_details_77341631]

再执行：

source /etc/bash.bashrc
    Ctrl+d  #(推迟root)
    sudo updatedb #更新database

（8）安装完成后通过查看 opencv 版本验证是否安装成功：

pkg-config --modversion opencv

附上测试摄像头的代码：

#!/usr/bin/python
    # -*- coding: UTF-8 -*-
    # 打开摄像头并灰度化显示
    import cv2
    
    capture = cv2.VideoCapture(0)
    
    while(True):
        # 获取一帧
        ret, frame = capture.read()
        # 将这帧转换为灰度图
        gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
    
        cv2.imshow('frame', gray)
        if cv2.waitKey(1) == ord('q'):
            break
    capture.release()
    cv2.destroyAllWindows()

测试实例2：

//文件名字lena.cpp
    #include <stdio.h>
    #include <opencv2/opencv.hpp>
    using namespace cv;
    
    int main( )
    {
        Mat image;
    
        //按照自己的目录，或者将
        image = imread("/home/slam2/catkin_ws/src/lena.png", 1 );
        if ( !image.data )
        {
            printf("No image data \n");
            return -1;
        }
        namedWindow("Display Image", WINDOW_AUTOSIZE );
        imshow("Display Image", image);
        waitKey(0);
        return 0;
    }

`Lena.png` 如下

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3podWlxaXV6aHVveXVlNTgz_size_16_color_FFFFFF_t_70 2][]

编译：

g++ lena.cpp -o lena.o  `pkg-config --cflags --libs opencv`

运行：

./lena.o

效果如下

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3podWlxaXV6aHVveXVlNTgz_size_16_color_FFFFFF_t_70 3][]

参考：[https://blog.csdn.net/yhaolpz/article/details/71375762\#t8][https_blog.csdn.net_yhaolpz_article_details_71375762_t8]

[https://blog.csdn.net/radiantjeral/article/details/82193370\#32\_OpenCV\_218][https_blog.csdn.net_radiantjeral_article_details_82193370_32_OpenCV_218]

安装CUDA10.2：[https://blog.csdn.net/zhuiqiuzhuoyue583/article/details/107356088][https_blog.csdn.net_zhuiqiuzhuoyue583_article_details_107356088]

[Link 1]: https://www.baidu.com/s?wd=%E9%98%89%E5%89%B2%E7%89%88&tn=24004469_oem_dg&rsv_dl=gh_pl_sl_csd
[http_opencv.org_releases.html]: http://opencv.org/releases.html
[SouthEast]: /images/20221124/ce80bbfec20944f18fd39f42c4affc2c.png
[SouthEast 1]: /images/20221124/b8f98815d1374161be140a14ccacc603.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3podWlxaXV6aHVveXVlNTgz_size_16_color_FFFFFF_t_70]: /images/20221124/e53abfcddda84cc18296ed2ccb1ea0fd.png
[https_blog.csdn.net_zhuiqiuzhuoyue583_article_details_106620707]: https://blog.csdn.net/zhuiqiuzhuoyue583/article/details/106620707
[https_blog.csdn.net_u014613745_article_details_78310916]: https://blog.csdn.net/u014613745/article/details/78310916
[https_blog.csdn.net_renhanchi_article_details_80493232]: https://blog.csdn.net/renhanchi/article/details/80493232
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3podWlxaXV6aHVveXVlNTgz_size_16_color_FFFFFF_t_70 1]: /images/20221124/742a65ae406944a9b5223e511a71f9ba.png
[https_blog.csdn.net_zhuiqiuzhuoyue583_article_details_106626618]: https://blog.csdn.net/zhuiqiuzhuoyue583/article/details/106626618
[https_blog.csdn.net_cv_you_article_details_77341631]: https://blog.csdn.net/cv_you/article/details/77341631
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3podWlxaXV6aHVveXVlNTgz_size_16_color_FFFFFF_t_70 2]: /images/20221124/ac6c3e44b50f4eb9bd0bc71951bb569b.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3podWlxaXV6aHVveXVlNTgz_size_16_color_FFFFFF_t_70 3]: /images/20221124/abacf153d8a04cb9a24c3c480082081d.png
[https_blog.csdn.net_yhaolpz_article_details_71375762_t8]: https://blog.csdn.net/yhaolpz/article/details/71375762#t8
[https_blog.csdn.net_radiantjeral_article_details_82193370_32_OpenCV_218]: https://blog.csdn.net/radiantjeral/article/details/82193370#32_OpenCV_218
[https_blog.csdn.net_zhuiqiuzhuoyue583_article_details_107356088]: https://blog.csdn.net/zhuiqiuzhuoyue583/article/details/107356088