opencv3 人脸识别 不同颜色圆圈 圈出



//原代码

#include <opencv2/highgui/highgui.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#include <opencv2/core/core.hpp>
#include <opencv2/objdetect/objdetect.hpp>

using namespace cv;
using namespace std;

void detectAndDraw(Mat& img, CascadeClassifier& cascade,
 CascadeClassifier& nestedCascade,
 double scale, bool tryflip);

int main()
{
 //VideoCapture cap(0);    //打开默认摄像头
 //if(!cap.isOpened())
 //{
 //    return -1;
 //}
 Mat frame;
 Mat edges;

 CascadeClassifier cascade, nestedCascade;
 bool stop = false;
 //训练好的文件名称，放置在可执行文件同目录下
 cascade.load("D:\\opencv\\sources\\data\\haarcascades\\haarcascade_frontalface_alt.xml");
 nestedCascade.load("D:\\opencv\\sources\\data\\haarcascades\\haarcascade_eye.xml");
 frame = imread("E:\\tmpimg\\hezhao.jpg");
 detectAndDraw(frame, cascade, nestedCascade, 2, 0);
 waitKey();
 /*
 while(!stop)
 {
    cap>>frame;
    detectAndDraw( frame, cascade, nestedCascade,2,0 );
    if(waitKey(30) >=0)
         stop = true;
 }
 */
 return 0;
}
void detectAndDraw(Mat& img, CascadeClassifier& cascade,
 CascadeClassifier& nestedCascade,
 double scale, bool tryflip)
{
 int i = 0;
 double t = 0;
 //建立用于存放人脸的向量容器
 vector<Rect> faces, faces2;
 //定义一些颜色，用来标示不同的人脸
 const static Scalar colors[] = {
  CV_RGB(0,0,255),
  CV_RGB(0,128,255),
  CV_RGB(0,255,255),
  CV_RGB(0,255,0),
  CV_RGB(255,128,0),
  CV_RGB(255,255,0),
  CV_RGB(255,0,0),
  CV_RGB(255,0,255) };
 //建立缩小的图片，加快检测速度
 //nt cvRound (double value) 对一个double型的数进行四舍五入，并返回一个整型数！
 Mat gray, smallImg(cvRound(img.rows / scale), cvRound(img.cols / scale), CV_8UC1);
 //转成灰度图像，Harr特征基于灰度图
 cvtColor(img, gray, CV_BGR2GRAY);
 imshow("灰度", gray);
 //改变图像大小，使用双线性差值
 resize(gray, smallImg, smallImg.size(), 0, 0, INTER_LINEAR);
 imshow("缩小尺寸", smallImg);
 //变换后的图像进行直方图均值化处理
 equalizeHist(smallImg, smallImg);
 imshow("直方图均值处理", smallImg);
 //程序开始和结束插入此函数获取时间，经过计算求得算法执行时间
 t = (double)cvGetTickCount();
 //检测人脸
 //detectMultiScale函数中smallImg表示的是要检测的输入图像为smallImg，faces表示检测到的人脸目标序列，1.1表示
 //每次图像尺寸减小的比例为1.1，2表示每一个目标至少要被检测到3次才算是真的目标(因为周围的像素和不同的窗口大
 //小都可以检测到人脸),CV_HAAR_SCALE_IMAGE表示不是缩放分类器来检测，而是缩放图像，Size(30, 30)为目标的
 //最小最大尺寸
 cascade.detectMultiScale(smallImg, faces,
  1.1, 2, 0
  //|CV_HAAR_FIND_BIGGEST_OBJECT
  //|CV_HAAR_DO_ROUGH_SEARCH
  | CV_HAAR_SCALE_IMAGE
  , Size(30, 30));
 //如果使能，翻转图像继续检测
 if (tryflip)
 {
  flip(smallImg, smallImg, 1);
  imshow("反转图像", smallImg);
  cascade.detectMultiScale(smallImg, faces2,
   1.1, 2, 0
   //|CV_HAAR_FIND_BIGGEST_OBJECT
   //|CV_HAAR_DO_ROUGH_SEARCH
   | CV_HAAR_SCALE_IMAGE
   , Size(30, 30));
  for (vector<Rect>::const_iterator r = faces2.begin(); r != faces2.end(); r++)
  {
   faces.push_back(Rect(smallImg.cols - r->x - r->width, r->y, r->width, r->height));
  }
 }
 t = (double)cvGetTickCount() - t;
 //   qDebug( "detection time = %g ms\n", t/((double)cvGetTickFrequency()*1000.) );
 for (vector<Rect>::const_iterator r = faces.begin(); r != faces.end(); r++, i++)
 {
  Mat smallImgROI;
  vector<Rect> nestedObjects;
  Point center;
  Scalar color = colors[i % 8];
  int radius;

  double aspect_ratio = (double)r->width / r->height;
  if (0.75 < aspect_ratio && aspect_ratio < 1.3)
  {
   //标示人脸时在缩小之前的图像上标示，所以这里根据缩放比例换算回去
   center.x = cvRound((r->x + r->width*0.5)*scale);
   center.y = cvRound((r->y + r->height*0.5)*scale);
   radius = cvRound((r->width + r->height)*0.25*scale);
   circle(img, center, radius, color, 3, 8, 0);
  }
  else
   rectangle(img, cvPoint(cvRound(r->x*scale), cvRound(r->y*scale)),
    cvPoint(cvRound((r->x + r->width - 1)*scale), cvRound((r->y + r->height - 1)*scale)),
    color, 3, 8, 0);
  if (nestedCascade.empty())
   continue;
  smallImgROI = smallImg(*r);
  //同样方法检测人眼
  nestedCascade.detectMultiScale(smallImgROI, nestedObjects,
   1.1, 2, 0
   //|CV_HAAR_FIND_BIGGEST_OBJECT
   //|CV_HAAR_DO_ROUGH_SEARCH
   //|CV_HAAR_DO_CANNY_PRUNING
   | CV_HAAR_SCALE_IMAGE
   , Size(30, 30));
  for (vector<Rect>::const_iterator nr = nestedObjects.begin(); nr != nestedObjects.end(); nr++)
  {
   center.x = cvRound((r->x + nr->x + nr->width*0.5)*scale);
   center.y = cvRound((r->y + nr->y + nr->height*0.5)*scale);
   radius = cvRound((nr->width + nr->height)*0.25*scale);
   circle(img, center, radius, color, 3, 8, 0);
  }
 }
 imshow("识别结果", img);
}

识别结果：