想使用cmake来组织代码工程，其中使用opencv与Qt，下面使用的CMakeLists.txt存档于此备忘。

opencv的使用例子

cmake_minimum_required(VERSION 3.15)
project(main)
set(CMAKE_CXX_STANDARD 11)
set(CMAKE_CXX_STANDARD_REQUIRED True)

message(STATUS "编译OpenCV...")
# aux_source_directory(. SOURCES)
# add_executable(main ${SOURCES})

find_path(INCLUDES opencv2/opencv.hpp /usr/local/include/opencv4/)

message(STATUS ${INCLUDES})

find_library(LIB1  opencv_core)
find_library(LIB2  opencv_imgcodecs)
find_library(LIB3  opencv_imgproc)
find_library(LIB4  opencv_ml)

include_directories(${INCLUDES})

add_executable(main main.cpp)
target_link_libraries(main ${LIB1} ${LIB2} ${LIB3} ${LIB4})

Qt5的使用例子

 cmake_minimum_required(VERSION 3.15)
project(main)
set(CMAKE_INCLUDE_CURRENT_DIR ON)  
set(CMAKE_CXX_STANDARD 11)
set(CMAKE_CXX_STANDARD_REQUIRED True)
SET(CMAKE_PREFIX_PATH /Users/yangqiang/Qt512/5.12.1/clang_64/lib/cmake)
find_package(Qt5Core REQUIRED)
FIND_PACKAGE(Qt5Widgets REQUIRED)
FIND_PACKAGE(Qt5Gui REQUIRED)
FIND_PACKAGE(Qt5Core REQUIRED)

SET(CMAKE_AUTOMOC ON)
SET(CMAKE_AUTOUIC ON)


qt5_wrap_ui(UI_FILES output.ui)


message(STATUS "编译Qt...")
aux_source_directory(. SOURCES)
add_executable(main ${SOURCES} ${UI_FILES})


include_directories(
    /Users/yangqiang/Qt512/5.12.1/Src/qtbase/include
    /Users/yangqiang/Qt512/5.12.1/Src/qtbase/include/QtWidgets
    /Users/yangqiang/Qt512/5.12.1/Src/qtbase/include/QtCore
    /Users/yangqiang/Qt512/5.12.1/Src/qtbase/include/GtGui
    /Users/yangqiang/Qt512/5.12.1/Src/qtbase/include/QtWidgets
    /Users/yangqiang/Qt512/5.12.1/Src/qtbase/include/QThread
)
find_library(FRAMEWORK1  QtCore /Users/yangqiang/Qt512/5.12.1/clang_64/lib)
find_library(FRAMEWORK2  QtGui /Users/yangqiang/Qt512/5.12.1/clang_64/lib)
find_library(FRAMEWORK3  QtWidgets /Users/yangqiang/Qt512/5.12.1/clang_64/lib)
target_link_libraries(main ${FRAMEWORK1} ${FRAMEWORK2} ${FRAMEWORK3})

• 项目截图

  
  image.png

opencv代码

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

void loadIrisFromTxt(cv::String filename, cv::Mat &data, cv::Mat &label){
    int rows = 150;
    int cols = 4;
    data.create(rows, cols, CV_32FC1);
    label.create(rows, 1, CV_32SC1);
    // 打开文件
    std::ifstream ifs(filename, std::ifstream::in);
    char buf[256];
    int row_id = 0;
    while(! ifs.eof()){
        bzero(buf, sizeof(buf));
        ifs.getline(buf, sizeof(buf)-1);
        std::istringstream istr(buf);
        float a, b, c, d;
        char buffer[256] = {0};
        char _;
        istr >> a >> _ >> b >> _ >> c >> _ >> d >> _ >> buffer;
        cv::Mat row_one = (cv::Mat_<float>(1,4) << a, b, c, d);
        row_one.copyTo(data.row(row_id));

        if(strcmp(buffer, "Iris-setosa") == 0){
            label.at<int>(row_id, 0) = 0;
        }
        else if (strcmp(buffer, "Iris-versicolor") == 0){
            label.at<int>(row_id, 0) = 1;
        }
        else if (! strcmp(buffer, "Iris-virginica")){
            label.at<int>(row_id, 0) = 2;
        }
        row_id ++;
    }
    // 关闭文件
    ifs.close();
};
// 可视化分类器与数据集
void visual_classifier( cv::Mat data,      // 可视化数据集，必须2两个特征
                        cv::Mat label,     // 可视化标签，必须一个特征
                        cv::Mat &out,       // 输出的可视化图
                        cv::Ptr<cv::ml::SVM> classifer = 0,  // 需要可视化的分类器，可以为0
                        int w=512,          // 默认可视化区域
                        int h=512){         // 默认可视化区域
    // 可视化数据集与分类器，其中数据集需要方大到指定w与h，标签用来区分样本点的颜色
    // out.create(h, w, CV_8UC3);  // 初始化输出的图像
    out = cv::Mat(h, w, CV_8UC3, cv::Scalar(255, 255, 255));
    // 计算缩放比例
    double scale_x, scale_y; 
    // 计算数据集的x，y最大最小值
    double max_x, min_x;
    double max_y, min_y;
    cv::minMaxIdx(data.col(0), &min_x, &max_x);
    cv::minMaxIdx(data.col(1), &min_y, &max_y);
    // 扩边，避免点绘制在边界上
    min_x -= 0.5;
    max_x += 0.5;
    min_y -= 0.5;
    max_y += 0.5;

    scale_x = w / (max_x - min_x);
    scale_y = h / (max_y - min_y);
    
    /////////////////绘制分类器的分类边界开始//////////////////
    std::cout << "开始绘制分类区域！"<< std::endl;
    // 定义绘制颜色
    cv::Vec3b cr(0, 0, 128), cg(0, 128, 0), cb(128, 0, 0);

    //对区域循环判别作为的类别
    for(int y = 0; y < h; y++){
        for(int x = 0; x < w; x++){
            // 把x，y映射到样本区域；
            // 先缩放
            double a_x = x / scale_x;
            double a_y = y / scale_y;
            // 再平移
            a_x += min_x;
            a_y += min_y;
            // 使用这个点预测
            cv::Mat a_sample = (cv::Mat_<float>(1,2) << a_x, a_y);
            double pre = classifer->predict(a_sample);
            if(pre == 0) {
                out.at<cv::Vec3b>(y, x) = cr;
            }
            else if(pre == 1 ){
                out.at<cv::Vec3b>(y, x) = cg;
            }else if(pre == 2 ){
                out.at<cv::Vec3b>(y, x) = cb;
            }
        }
    }
    /////////////////绘制分类器的分类边界结束///////////////////
    // 绘制支撑向量
    /////////////////绘制支撑向量开始///////////////////
    // 获取支撑向量
    cv::Mat sv = classifer->getSupportVectors();
    std::cout<< "支撑向量维度：" << sv.rows <<"," << sv.cols << std::endl;
    for (int i = 0; i < sv.rows; i++){
        const float* v = sv.ptr<float>(i);
        double v_x = v[0];
        double v_y = v[1];
        // 平移
        v_x -= min_x;
        v_y -= min_y;
        // 方大
        v_x *= scale_x;
        v_y *= scale_y; 
        // 绘制
        circle(out,  cv::Point( (int) v_x, (int) v_y), 3, cv::Scalar(255, 255, 255), 1);

    }
    /////////////////绘制支撑向量结束///////////////////
    std::cout << "开始绘制数据集样本点！"<< std::endl;
    // 定义绘制的颜色
    cv::Vec3b r(0, 0, 255), g(0, 255, 0), b(255, 0, 0);
    // 绘制样本点
    for(int i = 0; i < data.rows; i++){
        double d_x = data.at<float>(i, 0);
        double d_y = data.at<float>(i, 1);
        // 先平移到0点再放大
        d_x -= min_x;
        d_y -= min_y;
        d_x *= scale_x;
        d_y *= scale_y; 
        // 取整绘制
        int lab = label.at<int>(i,0);
        if(lab == 0){
            circle(out, cv::Point( (int) d_x,  (int) d_y ), 1, r, -1);
        }
        if(lab == 1){
            circle(out, cv::Point( (int) d_x,  (int) d_y ), 1, g, -1);
        }
        if(lab == 2){
            circle(out, cv::Point( (int) d_x,  (int) d_y ), 1, b, -1);
        }
    }
}

int main(int argc, const char** argv) {
    cv::Mat data, label;
    loadIrisFromTxt("iris.txt", data, label);

    cv::Mat newData;
    cv::hconcat(data.col(0), data.col(2), newData);

    // 创建SVM对象
    cv::Ptr<cv::ml::SVM>  svm = cv::ml::SVM::create();
    svm->setType(cv::ml::SVM::C_SVC);           // 带正则化惩罚因子C的SVM
    svm->setC(1000);                            // 设置C的值
    svm->setGamma(0.001);                        // 设置gamma的值
    svm->setKernel(cv::ml::SVM::SIGMOID);        // 设置SVM的核
    svm->setTermCriteria(cv::TermCriteria(cv::TermCriteria::MAX_ITER, (int)1e8, 1e-8));  // 设置训练终止条件
    // 训练终止条件
    svm->setTermCriteria(cv::TermCriteria(cv::TermCriteria(cv::TermCriteria::MAX_ITER, (int)1e8, 1e-8)));
    // 开始训练
    svm->train(newData,cv::ml::ROW_SAMPLE,label);
    // 使用训练集预测
    cv::Mat  predict;
    svm->predict(newData, predict);   // 第三个参数取默认值，预测返回的结果是CV_32F类型
    predict.convertTo(predict, CV_32SC1);   // 需要转换为一致才能比较

    // 统计准确率
    cv::Mat result;
    cv::compare(label, predict, result, cv::CMP_EQ); // 比较结果, 相等true，则设置为255，false设置为0
    cv::Scalar mean = cv::mean(result / 255.0);
    std::cout<<"识别准确率：";
    std::cout << std::fixed << std::setprecision(2) << mean[0] * 100.0 <<"%"<< std::endl;
    
    cv::Mat img;

    visual_classifier(newData, label, img, svm);
    imwrite("visual.png", img); 
    return 0;
}

• 输出图片

  
  SVM对鸢尾花的分类可视化