ubuntu16.04+anaconda2+opencv3.4.0

小咪咪 2022-04-12 09:10 235阅读 0赞

1.编译好的opencv和自己编译安装opencv的区别

Ubuntu安装了anaconda以后，

$conda install opencv

直接安装网上编译好的opencv，但是Ubuntu，python以及opencv不同版本之间使用都有差异，容易出问题；但是如果没出问题的话，这确实是最简单的安装方式.

$pip install opencv-python

或者$apt-get install python-opencv

或者$conda install –c https://conda.binstar.org/menpoopencv

具有同样问题，所以有的人选择自己编译安装opencv.

比如直接

$conda install opencv

之后，调用cv2.destroyAllWindows()时有时会出现：

Thefunction is not implemented. Rebuild the library with Windows, GTK+ 2.x orCarbon support. If you are on Ubuntu or Debian, install libgtk2.0‑dev and pkg‑config, then re‑runcmake or configure script

这就是网上编译好的opencv包在编译的过程中设置的选项可能和你当前需求不一样，解决方法就是自己编译安装opencv

2. Ubuntu下anaconda卸载opencv

安装前首先把之前的opencv卸载掉，conda install装的就用condauninstall卸，其他的安装方式也按照对应方式卸载就可以了.

然后

$cd /usr/local/lib

$sudo rm–r opencv相关的包

$cd/usr/local/include

$sudo rm–r opencv相关的包

看看/usr/local/下有没有其他opencv，有的话一并删除

删除一系列包可以用rm –r opencv\*，以opencv开头的包就都会被删掉

3.安装opencv依赖包

在安装opencv之前需要先处理好依赖包，有的教程会告诉你下面两条：

$sudo apt-get update

$sudo apt-get upgrade

这两条不是必要的，有的包升级完了又是bug，比如numpy包升到1.12就会出现floatX64的问题，所以不用这两条也没问题

下面这些依赖包还是要装的：

sudo apt-get install build-essential libgtk2.0-dev libjpeg-dev libtiff5-dev libjasper-dev libopenexr-dev cmake python-dev python-numpy python-tk libtbb-dev libeigen3-dev yasm libfaac-dev libopencore-amrnb-dev libopencore-amrwb-dev libtheora-dev libvorbis-dev libxvidcore-dev libx264-dev libqt4-dev libqt4-opengl-dev sphinx-common texlive-latex-extra libv4l-dev libdc1394-22-dev libavcodec-dev libavformat-dev libswscale-dev default-jdk ant libvtk5-qt4-dev

$   sudoapt-get install build-essential libgtk2.0-dev libjpeg-dev libtiff5-dev libjasper-dev libopenexr-dev cmake python-dev python-numpy python-tk libtbb-dev libeigen3-dev yasm libfaac-dev libopencore-amrnb-dev libopencore-amrwb-dev libtheora-dev libvorbis-dev libxvidcore-dev libx264-dev libqt4-dev libqt4-opengl-dev sphinx-common texlive-latex-extra libv4l-dev libdc1394-22-dev libavcodec-dev libavformat-dev libswscale-dev default-jdk ant libvtk5-qt4-dev

4.编译安装opencv

(1)官网下载opencv源码 解压opencv源码

(2)$cd 解压好的opencv文件夹

$mkdir build

$cd build

//万一编译失败，删掉build和/usr/local/下相应的opencv文件就行，不用把整个解压的opencv都删掉

(3)下载 ippicv\_2017u3\_lnx\_intel64\_general\_20170822.tgz链接:(可以不执行，若下载速度慢，再执行此步骤)

[https://github.com/opencv/opencv\_3rdparty/tree/ippicv/master\_20170822/ippicv][https_github.com_opencv_opencv_3rdparty_tree_ippicv_master_20170822_ippicv]

保存路径随意，我是放在默认的下载路径/home/lc/下载

修改opencv里相关配置文件:

打开终端，输入

gedit /home/lc/opencv\_source/opencv/3rdparty/ippicv/ippicv.cmake \#记得lc换成自己的用户名

将47行的

"https://raw.githubusercontent.com/opencv/opencv\_3rdparty/$\{IPPICV\_COMMIT\}/ippicv/"

改为步骤1中手动下载的文件的本地路径：

"file:///home/lc/下载/" \#（仅供参考，根据自己的路径填写）

编辑完成保存退出。

wanglin@wanglin:~/caffe_software/opencv-3.4.0/build$ cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local -D WITH_TBB=ON -D BUILD_NEW_PYTHON_SUPPORT=ON -D WITH_V4L=ON -D INSTALL_C_EXAMPLES=ON -D INSTALL_PYTHON_EXAMPLES=ON -D BUILD_EXAMPLES=ON -D WITH_OPENGL=ON -D WITH_VTK=ON ..

//opencv3.x版本这么写是没有问题的，2.x版本需要指定CUDA，而3.x版本指定CUDA的话编译会出问题.

//有的cmake里还有WITH\_QT=ON，QT4和QT5版本问题在这里也经常出现，这条别写了

//看看你的ffmpeg是否是YES，有的ffmpeg包在2.x版本是NO的状态，make不下去的，先去装好ffmpeg

wanglin@wanglin:~/caffe_software/opencv-3.4.0/build$ make

出现问题:

(1)注意不要在make后加-j16

(2)collect2: error: ld returned 1 exit status

modules/viz/CMakeFiles/opencv\_test\_viz.dir/build.make:236: recipe for target 'bin/opencv\_test\_viz' failed

make\[2\]: \*\*\* \[bin/opencv\_test\_viz\] Error 1

CMakeFiles/Makefile2:3306: recipe for target 'modules/viz/CMakeFiles/opencv\_test\_viz.dir/all' failed

make\[1\]: \*\*\* \[modules/viz/CMakeFiles/opencv\_test\_viz.dir/all\] Error 2

Makefile:160: recipe for target 'all' failed

make: \*\*\* \[all\] Error 2

*解决方式：*

sudo apt-get autoremove libtiff5-dev

sudo apt-get install libtiff5-dev

wanglin@wanglin:~/caffe_software/opencv-3.4.0/build$ sudo make install

将自己生成的动态链接库添加到系统能找到的路径:

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

在打开的写字板中填写/usr/local/lib

保存，并$sudo ldconfig

这个写法和下面的写法是等效的

$sudo -s

$echo"usr/local/lib" > /etc/ld.so.conf.d/opencv.conf

$sudoldconfig

9.设置路径：

$sudo mousepad  /etc/bash.bashrc

文件末尾添加并保存：

PKG\_CONFIG\_PATH=$PKG\_CONFIG\_PATH:/usr/local/lib/pkgconfig

export PKG\_CONFIG\_PATH

保存退出，然后执行如下命令使得配置生效

source /etc/bash.bashrc
    //激活配置然后更新database
    sudo updatedb

10.测试

$python

$import cv2

如果出现no module named cv2，那就把/home/wanglin/opencv-4.1.1/my\_build\_dir/lib/cv2.so拷到anaconda/lib/python2.7/site-packages下

11.一个demo

`import cv2`

`img = cv2.imread('1.jpg')`

`cv2.imshow('image',img)`

`cv2.waitKey(0)`

`cv2.destroyAllWindows()`

没有报错就表明安装成功.

[https_github.com_opencv_opencv_3rdparty_tree_ippicv_master_20170822_ippicv]: https://github.com/opencv/opencv_3rdparty/tree/ippicv/master_20170822/ippicv