opencv 移植

1.ubunut系统搭建opencv+python开发环境

1.1.ubuntu系统安装pip3工具

sudo apt-get install python3-pip //安装python模块安装工具pip3

sudo apt install python3-tk //安装tkinter模块(类似)，图形显示模块

1.2.打开pycharm开发工具，点击File->New Project->工程保存在/opt/project/opencv目录下

1.3.然后点击File->Setting->Project opencv->Project interpreter->右侧：Project interpreter：Python3.6 /usr/bin/python3.6

1.4.然后点击“+”号，在弹出的对话框中输入“opencv”进行搜索,将"opencv-python"和"opencv-contrib-python"分别选中然后点击“Install Package”安装即可

同样搜索numpy，matplotlib安装

1.5.拷贝ftp://project/code/day15_day16/images图片目录到/opt/project/opencv/目录下

cp images /opt/project/opencv/

1.6.然后新建一个文件helloopencv.py测试是否能够显示图片，支持开发环境搭建完毕

工程保存在:/opt/project/opencv/opencv_test

添加代码：

import cv2 as cv

src = cv.imread("/opt/project/opencv/images/lena.png")

cv.imshow("input", src)

cv.waitKey(0)

cv.destroyAllWindows()

1.7.其余代码参见opencv目录代码和验证即可

2.ubuntu系统搭建opencv,c++开发环境

sudo apt-get install libopencv-dev

测试：

cd /opt/project/opencv/opencv_test

vim opencv_test.cpp 添加

#include

#include <opencv2/opencv.hpp>

using namespace std;
using namespace cv;

int main()
{
	Mat srcImage = imread("/opt/project/opencv/images/lena.png");
	imshow("源图像",srcImage);

	waitKey(0);

	return 0;

}

保存推出

g++ pkg-config opencv --cflags opencv_test.cpp -o opencv_test pkg-config opencv --libs

./opencv_test

2.opencv移植步骤[目前支持C++]：

上位机执行：

2.1.获取源码：ftp://project/code/day14/opencv-3.4.3.zip

2.2.安装：sudo apt-get install cmake-qt-gui

2.3.配置编译opencv

mkdir /opt/project/opencv_source

mkdir /opt/project/opencv_source/opencv_arm

mkdir /opt/project/opencv_source/opencv_install

cp opencv-3.4.3.zip /opt/project/opencv_source/

cd /opt/project/opencv_source

unzip opencv-3.4.3.zip

cd /opt/project/opencv/opencv-3.4.3

cmake-gui

然后在出现的界面中做一下配置：

1.选择源代码目录:/opt/project/opencv_source/opencv-3.4.3/

2.选择Build目录:/opt/project/opencv_source/opencv_arm

3.点击Configure，保持generator为Unix Makefiles，选择Specify options for cross-compiling，点击Next

4.Operating System填写Linux

5.C Compilers填写/opt/toolchains/bin/arm-cortex_a9-linux-gnueabi-gcc

6.C++ Compilers填写/opt/toolchains/bin/arm-cortex_a9-linux-gnueabi-g++

7.程序库的Target Root填写/opt/toolchains/include

8.点击Finish

9.修改默认配置，默认安装目录为/usr/local，对于交叉编译的库来说并不合适，所以我把CMAKE_INSTALL_PREFIX变量改为/opt/project/opencv_source/opencv_install

10.选中INSTALL_PYTHON_EXAMPLE

11.将PYTHON3_EXECUTABLE修改为自己交叉编译的python路经:/opt/project/python_arm_install/bin/python3

12.将PYTHON3_INCLUDE_DIR修改为自己交叉编译的python路径:/opt/project/python_arm_install//include/python3.5m

13.将PYTHON3_LIBRARY修改为自己交叉编译python路径：/opt/project/python_arm_install/lib/python3.5/config-3.5m/libpython3.5m.a

//14.将PYTHON3_NUMPY_INCLUDE_DIRS修改为/usr/local/lib/python3.6/dist-packages/numpy/core/include

然后选择WITH_LIBV4L和WITH_V4L和WITH_QT

然后点击Configure

然后再次修改QT相关选项：

Qt5Concurrent_DIR:PATH=/opt/project/qt/lib/cmake/Qt5Concurrent

Qt5Core_DIR:PATH=/opt/project/qt/lib/cmake/Qt5Core

Qt5Gui_DIR:PATH=/opt/project/qt/lib/cmake/Qt5Gui

Qt5Test_DIR:PATH=/opt/project/qt/lib/cmake/Qt5Test

Qt5Widgets_DIR:PATH=/opt/project/qt/lib/cmake/Qt5Widgets

15.然后按Configure再点击Genertor

2.4.修改配置：

cd /opt/project/opencv_source/opencv_arm 执行

find ./ -name "flags.make" -exec sed -i "s/CXX_FLAGS = -fsigned-char/CXX_FLAGS = -fpic -fsigned-char/g" {} ;

find ./ -name "flags.make" -exec sed -i "s/C_FLAGS = -fsigned-char/C_FLAGS = -fpic -fsigned-char/g" {} ;

vim CMakeCache.txt

将:PYTHON3_INCLUDE_PATH:INTERNAL=/usr/local/include/python3.5m

修改为：

PYTHON3_INCLUDE_PATH:INTERNAL=/opt/project/python_arm_install/include/python3.5m

vim CMakeCache.txt

将PYTHON3_LIBRARIES:INTERNAL=/usr/local/lib/libpython3.5m.a

修改为PYTHON3_LIBRARIES:INTERNAL=/opt/project/python3_5_6_install/lib/python3.5/config-3.5m/libpython3.5m.a

保存推出

vim /opt/project/opencv_source/opencv-3.4.3/modules/videoio/src/cap_v4l.cpp

将#define MAX_CAMERAS 8

修改为#define MAX_CAMERAS 9

保存推出

编译安装

make -j4

make install

mkdir /opt/rootfs/home/opencv/

cp /opt/project/opencv_source/opencv_install/lib /opt/rootfs/home/opencv/ -frd

cp /opt/project/opencv_source/opencv_install/share /opt/rootfs/home/opencv/ -frd

cp /opt/project/opencv_source/opencv_install/bin /opt/rootfs/home/opencv/ -frd

vim /opt/rootfs/etc/profile 文件最后添加

export LD_LIBRARY_PATH=/home/opencv/lib:$LD_LIBRARY_PATH

保存退出

重启下位机

2.5.opencv C++测试：

参考代码位于：ftp://project/code/day14/

1.上位机执行：

mkdir /opt/project/opencv_test/capture/

cd /opt/project/opencv_test/capture

vim capture.cpp //拍照程序

/opt/project/qt/bin/qmake -project

vim capture.pro 添加：

INCLUDEPATH+=/opt/project/opencv_source/opencv_install/include

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_core.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_highgui.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_calib3d.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_features2d.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_flann.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_imgproc.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_imgcodecs.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_ml.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_objdetect.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_photo.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_superres.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_shape.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_videoio.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_video.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_videostab.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_stitching.so

保存退出

/opt/project/qt/bin/qmake

make

cp caputure /opt/rootfs/home/apptest

mkdir /opt/project/opencv_test/video/

cd /opt/project/opencv_test/video/

vim video_stream.cpp //视频显示程序

vim video.pro 添加：

INCLUDEPATH+=/opt/project/opencv_source/opencv_install/include

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_core.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_highgui.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_calib3d.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_features2d.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_flann.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_imgproc.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_imgcodecs.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_ml.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_objdetect.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_photo.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_superres.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_shape.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_videoio.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_video.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_videostab.so

LIBS+=/opt/project/opencv_source/opencv_install/lib/libopencv_stitching.so

保存退出

/opt/project/qt/bin/qmake

make

cp video /opt/rootfs/home/apptest

2.下位机然后插入摄像头运行：

cd /home/apptest

./capture

运行提示各种动态库找不到，在上位机上从交叉编译器中拷贝即可：

cp /opt/toolchains/arm-cortex_a9-linux-gnueabi/sysroot/lib/libstdc++.so.6* /opt/rootfs/lib/ -frd

cp /opt/toolchains/arm-cortex_a9-linux-gnueabi/sysroot/lib/librt* /opt/rootfs/lib/ -frd

3.然后下位机执行：

cd /home/apptest

./capture

查看picture.jpg照片

./video //查看LCD显示的视频

4.尝试将参考代码中的camerface.cpp人脸识别的代码在下位机

运行,摄像头对准头像实现人脸检测！

2.6.然后django添加拍照显示功能！

0.添加摄像头拍照硬件操作库

mkdir /opt/project/hwlib/capture/

cd /opt/project/hwlib/capture/

vim capture.cpp 添加：

#include<opencv2/opencv.hpp>
  #include<opencv2/highgui/highgui.hpp>
  #include<opencv2/imgproc/imgproc.hpp>
  #include <opencv2/core/core.hpp>
  #include<iostream>
  #include<stdio.h>
  using namespace cv;
  using namespace std;

  extern "C"
  int camera(void)
  {
    VideoCapture  capture(9);
    Mat frame;
    char filename[200];
    capture >> frame;
    sprintf(filename, "/home/django/ehome/ehome/static/images/picture.jpeg");
    imwrite(filename, frame);
    return 0;
  }

保存退出

arm-cortex_a9-linux-gnueabi-g++ -shared -fpic -o libcapture.so capture.cpp

-I /opt/project/opencv_source/opencv_install/include/

-L /opt/project/opencv_source/opencv_install/lib/*.so

cp libcapture.so /opt/rootfs/home/applib

//注意一下操作步骤用pycharm工具搞定

1.修改urls.py,添加：

url('^capture$', view.capture),

2.修改view.py,文件最后添加：

拍照片

from ctypes import *

import os, sys

handle = CDLL('/home/applib/libcapture.so')

def capture(reqeuest):

ret = handle.camera()

if ret == 0:

return HttpResponse('拍照完毕，在主页面请刷新')

else:

return HttpResponse('拍照失败')

3.修改ehome.html,文件最后添加：

       <form action="/capture">
        <img id="picture" src="/static/images/picture.jpeg" width="320" height="240">
        <input style="color: dodgerblue " type="submit" value="点击拍照">
        <button onclick="reflush();return false">刷新</button>
        <script type="text/javascript">
            function reflush()
            {
                document.getElementById('picture').src="/static/images/picture.jpeg?"+Math.random();
                console.log("刷新")
            }
        </script>
    </form>
    <hr/>
    <br/>
    <br/>
    <br/>
   注意路径问题

4.修改settings.py文件，文件最后添加：

cd /opt/rootfs/home/django/ehome/ehome

vim settings.py 文件最后添加：

设置图片等静态文件的路径

STATIC_ROOT = os.path.join(os.path.dirname(file),'static')

STATICFILES_DIRS = (

('css',os.path.join(STATIC_ROOT,'css').replace('\','/') ),

('js',os.path.join(STATIC_ROOT,'js').replace('\','/') ),

('images',os.path.join(STATIC_ROOT,'images').replace('\','/') ),

('upload',os.path.join(STATIC_ROOT,'upload').replace('\','/') ),

)

5.创建目录

cd /opt/rootfs/home/django/ehome/ehome //注意路经问题

mkdir static

cd static

mkdir images css js

说明：

images:保存图片

css:保存CSS配置文件

js：保存JS文件

6.浏览器测试：192.168.1.110:8000/ehome

下位机提前启动服务器：python manage.py runserver 0.0.0.0:8000