博客

特征点检测与匹配算法——surf

#include <opencv2/opencv.hpp>
#include <opencv2/nonfree/nonfree.hpp>
#include <sys/time.h>
#include <iostream>

using namespace std;
int main()
{
    cv::Mat srcImage1=cv::imread("cloned1.png");
    cv::Mat srcImage2=cv::imread("cloned2.png");
   
    struct timeval t1,t2;
    struct timeval c1,c2;
    gettimeofday(&t1,NULL);
    
    int minHessian=400;//SURF算法中的hessian阈值
    cv::SurfFeatureDetector detector(minHessian);
    std::vector<cv::KeyPoint>keyPoints_object,keyPoints_scene;
    //调用detect函数检测出SURF特征关键点
    detector.detect(srcImage1, keyPoints_object);
    detector.detect(srcImage2, keyPoints_scene);

    gettimeofday(&c1,NULL);
    
    //计算描述符（特征向量）
    cv::SurfDescriptorExtractor extractor;
    cv::Mat descriptor_object,descriptor_scene;
    extractor.compute(srcImage1, keyPoints_object, descriptor_object);
    extractor.compute(srcImage2, keyPoints_scene, descriptor_scene);
    
    //使用FLANN匹配算子进行匹配
    cv::FlannBasedMatcher matcher;
    std::vector<cv::DMatch>matches;
    matcher.match(descriptor_object, descriptor_scene, matches);

    double max_dist=0;double min_dist=100;//最小距离和最大距离

    //计算出关键点之间距离的最大值和最小值
    for(int i=0;i<descriptor_object.rows;i++)
    {
        double dist=matches[i].distance;
        if(dist<min_dist)min_dist=dist;
        if(dist>max_dist)max_dist=dist;
    }
    //printf(">Max dist最大距离 ： %f \n",max_dist);
    //printf(">Min dist最小距离 ： %f \n",min_dist);

    //存下匹配距离小于3*min_dist的点对
    std::vector<cv::DMatch> good_matches;
    for(int i=0;i<descriptor_object.rows;++i)
    {
        if(matches[i].distance<3*min_dist)
        {
            good_matches.push_back(matches[i]);
        }
    }
    //绘制出匹配到的关键点
    cv::Mat img_matches;
            
 cv::drawMatches(srcImage1,keyPoints_object,srcImage2,keyPoints_scene,good_matches,img_matches,cv::Scalar::all(-1),cv::Scalar::all(-1),std::vector<char>(),cv::DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS);
    //定义两个局部变量
    std::vector<cv::Point2f> obj;
    std::vector<cv::Point2f> scene;

    //从匹配成功的匹配对中获取关键点
    for(unsigned int i=0;i<good_matches.size();++i)
    {
        obj.push_back(keyPoints_object[good_matches[i].trainIdx].pt);
        scene.push_back(keyPoints_scene[good_matches[i].queryIdx].pt);
    }
    cv::Mat H=cv::findHomography(obj,scene,CV_RANSAC,3,cv::noArray());//计算透视变换
    
    //从待测图片中获取角点
    std::vector<cv::Point2f> obj_corners(4);
    obj_corners[0]=cvPoint(0,0);obj_corners[1]=cvPoint(srcImage1.cols,0);
    obj_corners[2]=cvPoint(srcImage1.cols, srcImage1.rows);
    obj_corners[3]=cvPoint(0,srcImage1.rows);
    std::vector<cv::Point2f> scene_corners(4);

    //进行透视变换
    cv::perspectiveTransform(obj_corners, scene_corners, H);

    //绘制出角点之间的直线
    line(img_matches,scene_corners[0]+cv::Point2f(static_cast<float>(srcImage1.cols),0),scene_corners[1]+cv::Point2f(static_cast<float>(srcImage1.cols),0),cv::Scalar(255,0,123),4);
    line(img_matches,scene_corners[1]+cv::Point2f(static_cast<float>(srcImage1.cols),0),scene_corners[2]+cv::Point2f(static_cast<float>(srcImage1.cols),0),cv::Scalar(255,0,123),4);
    line(img_matches,scene_corners[2]+cv::Point2f(static_cast<float>(srcImage1.cols),0),scene_corners[3]+cv::Point2f(static_cast<float>(srcImage1.cols),0),cv::Scalar(255,0,123),4);
    line(img_matches,scene_corners[3]+cv::Point2f(static_cast<float>(srcImage1.cols),0),scene_corners[0]+cv::Point2f(static_cast<float>(srcImage1.cols),0),cv::Scalar(255,0,123),4);
   
    gettimeofday(&t2,NULL);
    cout<<t2.tv_usec-t1.tv_usec<<endl;
    cout<<t2.tv_usec-c1.tv_usec<<endl;
    imshow("Good Matches & Object detection",img_matches);

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