opencv计数项目实现及步骤

您好，这个图片计数可以实现吗？有什么难点，大致什么思路，可以说一下嘛？

opencv计数项目实现代码

#include<iostream>
#include<opencv.hpp>
using namespace std;
using namespace cv;
 
int main()
{
    //加载图像
    Mat src = imread("4.jpg");
    if (src.empty())
    {
        cout << "no image!" << endl;
        return -1;
    }
    imshow("src", src);
    //二值化
    Mat gray, binary;
    cvtColor(src, gray, COLOR_BGR2GRAY);
    imshow("gray", gray);
    threshold(gray, binary, 0, 255, THRESH_TRIANGLE);
    imshow("binary", binary);
    //图像形态学操作
    Mat kern1 = getStructuringElement(MORPH_RECT, Size(5, 5), Point(-1,-1));
    morphologyEx(binary, binary, MORPH_DILATE, kern1,Point(-1,-1),3);
    imshow("dilate", binary);
    //距离变换
    Mat dist;
    bitwise_not(binary, binary);//将白色背景变黑
    imshow("not", binary);
    distanceTransform(binary, dist, DIST_L2, 3);
    normalize(dist, dist, 0.0, 1.0, NORM_MINMAX);
    dist.convertTo(dist, CV_8UC1);
    imshow("dist", dist);
    //阈值分割
    Mat img;
    //threshold(dist, img, 0, 255, THRESH_BINARY | THRESH_OTSU);
    adaptiveThreshold(dist, img, 255, ADAPTIVE_THRESH_GAUSSIAN_C, THRESH_BINARY, 85, 0.0);
    imshow("img", img);
    //腐蚀操作
    morphologyEx(img, img, MORPH_DILATE, kern1, Point(-1,-1),1);
    imshow("dilate-img", img);
    //寻找连通域
    vector<vector<Point>>contours;
    findContours(img, contours, RETR_EXTERNAL, CHAIN_APPROX_SIMPLE);
    Mat mask = Mat::zeros(src.size(), CV_8UC3);
    for (size_t i = 0; i < contours.size(); i++)
    {
        drawContours(mask, contours, static_cast<int>(i), Scalar(0, 0, 255), -1, 8);
    }
    cout << "个数：" << contours.size() << endl;
    imshow("result", mask);
 
    waitKey(0);
    return 0;
}

这个和人眼识别是一样的，需要先确认检测物品然后计数，将已识别物品标记去掉，防止循环识别，将整张图过滤完后就可以输出计数了