User OpenCV to add watermark on a video¶

安装开发环境¶

首先，需要OpenCV开发环境，就是安装OpenCV相关的库。

直接到GitHub上拉下OpenCV源代码，然后进入项目文档中寻找从源代码编译的方法。我的平台是MacOS，所以找到的是Linux/Mac平台下的安装方法: http://docs.opencv.org/master/d7/d9f/tutorial_linux_install.html

但是参考了另一篇文章，下载了源代码并进入目录(opencv-master/)下，使以下命令编译:

sudo cmake -G "Unix Makefiles" //检查相关配置选项
sudo make -j8  //八线程编译
sudo make install  //安装编译好的库文件和头文件到系统目录下

测试开发环境是否安装成功¶

因为我的功能也不是很复杂，所以打算只用g++编译运行，就不动用各种大型IDE了。下在是我的测试用例：

#include <iostream>
#include <opencv2/opencv.hpp>

using namespace std;
using namespace cv;

int main(int argc, char **argv)
{
        Mat img = imread(argv[1]);
        imshow("test",img);
        waitKey(0);
        return 0;
}

编译使用：

sudo g++ opencv_test.cpp -lopencv_core -lopencv_highgui -lopencv_imgcodecs -I/usr/local/include -L/usr/local/lib

使用下面命令显示一幅图片:

./a.out test.jpg # test.jpg is your own picture file

如果能够显示出图片，说明开发环境已经安装成功;-)

读视频¶

废话不多话，直接上代码：

#include <iostream>
#include <sstream>
#include <opencv2/opencv.hpp>

using namespace std;
using namespace cv;

VideoWriter output;

Mat addWatermarkInFrame(Mat frame, int frameNum)
{
    cout << "processing " <<frameNum << " frame" <<endl; 

    Mat redSquare = Mat(Size(20,20),CV_8UC3,Scalar(0,0,255));

    Mat imageROI = frame(Rect(10,10,redSquare.cols, redSquare.rows));

    addWeighted(imageROI, 1.0, redSquare, 1.0,0,imageROI);  

    return frame;
}

void Stop(int signal)
{   
    output.~VideoWriter();
    cout << "process been interrupted!" << endl;
    exit(0);
}

void help()
{
    cout
    << "-----------------------------------------------------" << endl
    << "This program shows how to read a video file with OpenCV." << endl
    << "In addition, it show the process of add watermark in a video" << endl
    << "Usage:" << endl
    << "./a.out <input_video_file or video_stream_url>" << endl;
}

int main(int argc, char **argv)
{
    if(argc < 2)
    {
        help();
    }
    else if(argc > 2)
    {
        cout << "Too many parameters" << endl;
    }

    const string cameraOrFile= argv[1];

    stringstream ss;
    ss << cameraOrFile;

    int deviceID;
    ss >> deviceID;

    VideoCapture video;
    if(ss.good())
    {
        video.open(deviceID);
    }
    else
    {
        video.open(cameraOrFile);
    }

    if(!video.isOpened())
    {
        cout<< "Could not open the video file!" << endl;
        return -1;
    }

    Size videoSize = Size((int)video.get(CAP_PROP_FRAME_WIDTH),
                        (int)video.get(CAP_PROP_FRAME_HEIGHT));
    int frameRate = video.get(CAP_PROP_FPS);

    cout << "The size of video frame is: "<< videoSize << endl;
    cout << "The frame rate is: " << frameRate << endl; 

    int frameNum = -1;
    Mat frame;
    namedWindow("frame",WINDOW_AUTOSIZE);

    remove("output.mp4");
    output.open("output.mp4", -1, frameRate,videoSize,true);
    if(!output.isOpened())
    {
        cout << "Could not open the output video for write!" << endl;
        return -1;
    }

    signal(SIGINT,Stop);

    while(1)
    {
        video >> frame;
        if(frame.empty())
        {
            cout << "The video had been processed!" << endl;
            break;
        }

        Mat ret = addWatermarkInFrame(frame, ++frameNum);

        imshow("frame",ret);
        waitKey(1.0/frameRate * 1000);

        output << ret;
    }

    cout << "Finished writing video" << endl;

    return 0;
}

编译命令如下：

sudo g++ opencv_test.cpp -lopencv_core -lopencv_highgui -lopencv_imgcodecs -I/usr/local/include -L/usr/local/lib -lopencv_videoio

编译出的可执行文件是a.out, 使用方法可以是：

./a.out test.mp4 # process the video file

或者

./a.out "http://zhulongyixian.vicp.cc:30236/?action=stream" # <video_stream_url> process the video stream

不过如果原来的视频有声音的话，由于OpenCV处理后只有视频数据被写入了文件，这样造成处理结果文件中丢失了音频通道的数据。下一步需要考虑使用FFmpeg来改写视频编解码部分了。

加水印¶

加水印部分可以在上面程序的第10行开始的函数里面进行，为了简化处理，只给视频左上角加了一个红色小方块。

后记¶

最近学习linux相关知识，发现命令行越来越好用了，当然只是针对简单点的开发任务。