C++ opencv获取精确轮廓坐标

注释部分均可以运行，大概是画二值化，灰度化，所有轮廓的图像，画的图像太多了我就注视了，算精确坐标的部分就一个for循环，别的方法请告诉我，感谢！效果图放下边了

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


void show(std::string img_name, cv::Mat img) {
    cv::namedWindow(img_name, cv::WINDOW_NORMAL);
    cv::resizeWindow(img_name, 1000, 1000);
    cv::imshow(img_name, img);
}

int dis(int cx, int cy, int tx, int ty) {
    return pow(pow(cx - tx, 2) + pow(cy - ty, 2), 0.5);

}


int find_contour(cv::Mat img) {//找出工作票的轮廓

    cv::Mat img_rect;   // 绘制精确轮廓
    img.copyTo(img_rect);

    cv::Mat img_gray, img_binary;
    std::vector<std::vector<cv::Point>> contours;   //所有轮廓的集合
    std::vector<std::vector<cv::Point>> contours_select;   //筛选后轮廓的集合

    //转换为灰度图
    cv::cvtColor(img, img_gray, cv::COLOR_BGR2GRAY);

    // 二值化
    adaptiveThreshold(~img_gray, img_binary, 255,
                      cv::ADAPTIVE_THRESH_GAUSSIAN_C, cv::THRESH_BINARY, 15, -10);

    // 边缘检测
    cv::findContours(img_binary, contours, cv::RETR_LIST, cv::CHAIN_APPROX_NONE);

    // 筛选出面积最大的轮廓面积
    int maxrec_index = 0;   //最大面积外接矩形的index
    int maxarea = -1;
    for (int i = 0; i < contours.size(); ++i) {
        cv::RotatedRect rect = cv::minAreaRect(contours[i]);
        double w = rect.size.width;
        double h = rect.size.height;
        if (w * h > maxarea) {
            maxarea = w * h;
            maxrec_index = i;
        }
    }

    contours_select.push_back(contours[maxrec_index]);
//    cv::Mat img_maxrec;
//    img.copyTo(img_maxrec);   //深拷贝
//    cv::drawContours(img_maxrec, contours_select, -1,
//                     cv::Scalar(0, 255, 0), 2);    //  绘制最小矩形轮廓

    //绘制最小外接矩形
//    cv::Point2f vertices[4];
//    cv::RotatedRect rect = minAreaRect(contours[maxrec_index]);
//    rect.points(vertices);
//    std::cout << rect.center;
//    cv::Mat img_minrec;
//    img.copyTo(img_minrec);   //深拷贝
//    for (int i = 0; i < 4; ++i) {
//        line(img_minrec, vertices[i], vertices[(i + 1) % 4], cv::Scalar(0, 0, 255), 2);
//    }
//    cv::Rect brect = rect.boundingRect();
//    std::cout << brect;
//    cv::rectangle(img, brect, cv::Scalar(255, 0, 0), 2);

    int sumx = 0;
    int sumy = 0;
    for (int i = 0; i < contours_select[0].size(); ++i) {
        sumx += contours_select[0][i].x;
        sumy += contours_select[0][i].y;
    }
    int centerx = sumx / contours_select[0].size();
    int centery = sumy / contours_select[0].size();

//    std::vector<std::vector<cv::Point>> approx(contours.size());//拟合直线并绘制直线
//    for (int i = 0; i < contours.size(); i++) {
//        approxPolyDP(cv::Mat(contours[i]), approx[i], 15, true);
//
//        drawContours(img, approx, i, cv::Scalar(0, 255, 255), 2, 8);  //绘制
//    }


    // 找出精确工作票轮廓
    std::vector<cv::Point> cor(4);
    int maxdis[4] = {0,0,0,0};
    for (int i = 0; i < contours_select[0].size(); i++) {

        if (contours_select[0][i].x < centerx && contours_select[0][i].y < centery) { // left top point
            if (dis(centerx, centery, contours_select[0][i].x, contours_select[0][i].y) > maxdis[0]) {
                maxdis[0] = dis(centerx, centery, contours_select[0][i].x, contours_select[0][i].y);
                cor[0].x = contours_select[0][i].x;
                cor[0].y = contours_select[0][i].y;
            }
        } else if (contours_select[0][i].x > centerx && contours_select[0][i].y < centery) {  //  right top point
            if (dis(centerx, centery, contours_select[0][i].x, contours_select[0][i].y) > maxdis[1]) {
                maxdis[1] = dis(centerx, centery, contours_select[0][i].x, contours_select[0][i].y);
                cor[1].x = contours_select[0][i].x;
                cor[1].y = contours_select[0][i].y;
            }
        } else if (contours_select[0][i].x > centerx && contours_select[0][i].y > centery) {  //  right down point
            if (dis(centerx, centery, contours_select[0][i].x, contours_select[0][i].y) > maxdis[2]) {
                maxdis[2] = dis(centerx, centery, contours_select[0][i].x, contours_select[0][i].y);
                cor[2].x = contours_select[0][i].x;
                cor[2].y = contours_select[0][i].y;
            }
        } else if (contours_select[0][i].x < centerx && contours_select[0][i].y > centery) {  //  left down point
            if (dis(centerx, centery, contours_select[0][i].x, contours_select[0][i].y) > maxdis[3]) {
                maxdis[3] = dis(centerx, centery, contours_select[0][i].x, contours_select[0][i].y);
                cor[3].x = contours_select[0][i].x;
                cor[3].y = contours_select[0][i].y;
            }
        }
    }

    for (int i = 0; i < 4; ++i) {
        line(img_rect, cor[i], cor[(i + 1) % 4], cv::Scalar(0, 0, 255), 2);
    }

    show("img", img);
    show("img_binary", img_binary);
//    show("img_maxrec", img_maxrec);
//    show("img_minrec", img_minrec);
    show("img_rect", img_rect);


    cv::waitKey(0);
    return 1;
}

int main(){
    cv::Mat img = cv::imread("../photo/t2.jpg");
    find_contour(img);
    return 1;
}