OpenCV识别人的黄色皮肤并测出相机到人的距离

识别黄色的物体受光照影响大，hsv的值可以自己调，用d=f*w/p这条公式求距离误差有点大，可上网找其他的方法。

代码如下：

#include
#include
using namespace std;
using namespace cv;
int main()
{	
	VideoCapture cap;
	cap.open(1);   //打开相机 
	
	int H_min =2;
	int H_max = 15;
	int S_min = 44;
	int S_max = 255;
	int V_min = 91;
	int V_max = 255;

	Mat temp = Mat::zeros(Size(200, 200), CV_8UC3);   //调参窗口的图像 
	namedWindow("调参窗口", WINDOW_FREERATIO);
	namedWindow("识别", WINDOW_FREERATIO);
	namedWindow("二值", WINDOW_FREERATIO);
		
	Mat src, dst, stream;
	int dist;
	while (1)
	{
		cap >> src;
	
		blur(src, src, Size(3, 3));   //模糊降噪 
		cvtColor(src, dst, COLOR_BGR2HSV);   //转HSV格式 
		createTrackbar("H_min", "调参窗口", &H_min, 255);  //调节HSV的拖动条 
		createTrackbar("H_max", "调参窗口", &H_max, 255);
		createTrackbar("S_min", "调参窗口", &S_min, 255);
		createTrackbar("S_max", "调参窗口", &S_max, 255);
		createTrackbar("V_min", "调参窗口", &V_min, 255);
		createTrackbar("V_max", "调参窗口", &V_max, 255);

		inRange(dst, Scalar(H_min, S_min, V_min), Scalar(H_max, S_max, V_max), dst);   //转二值 
		Mat kernel = getStructuringElement(MORPH_RECT, Size(5, 5));  //结构元素 
		morphologyEx(dst, dst, MORPH_OPEN, kernel);  //开操作 ——去除杂质 
		morphologyEx(dst, dst, MORPH_CLOSE, kernel);  //闭操作 ——填充小洞 
		imshow("二值", dst);

		Canny(dst, dst, 100, 200);   //Canny边缘检测 
		vector>v;	      //点容器——用于存放轮廓			
		findContours(dst, v,  RETR_TREE, CHAIN_APPROX_NONE);   //发现轮廓	
		vector> couter(v.size());

		Rect rect;
		for (size_t i = 0; i < v.size(); i++)
		{
			if (contourArea(v[i]) > 300 && contourArea(v[i]) < 30000) //筛选轮廓面积 
			{ 
				approxPolyDP(v[i], couter[i], 10, true);

				if (couter[i].size() != 4 )  //筛选规则形状 
				{
					rect = boundingRect(v[i]);
					double biw = rect.width / rect.height;
					if (rect.area() > 300 && rect.area() < 30000 && biw>0.7 &&biw<2.0) //筛选矩形框面积和矩形宽与高的比值 
					{
						rectangle(src, rect, Scalar(0, 0, 255), 2, 8);  //画矩形 
						double x = 0, y = 0;
						x = (rect.br().x + rect.tl().x) / 2;
						y = (rect.br().y + rect.tl().y) / 2;
						circle(src, Point(x, y), 3, Scalar(255, 0, 0), -1, 8);//画出矩形中心点 

						/* 已知相机的焦距f,求相机到物体的距离 */
                        // f=p*d/w，先用 像素宽度*实际距离/物体实际宽度  求出相机焦距 
						//d=f*w/p	 再用  	焦距*实际宽度/像素宽度   求距离			
						double w = 27;		//实际宽度
						double f = 770;      //相机焦距 
						double p;			//像素宽度
						double d = 0;		//实际距离
						p = rect.width;
						d = f * w / p;

						dist = (int)d;
						String str = "distance:" + to_string(dist) + "cm";  
						putText(src, str, rect.br(), FONT_HERSHEY_PLAIN, 1, Scalar(0, 213, 255), 1, 8);//标出距离 
					
					}
				}
			}
		}
		imshow("识别", src);
		imshow("调参窗口", temp);

		waitKey(1);
	}

	return 0;
}