淺析ORB、SURF、SIFT特征點提取方法以及ICP匹配方法

Posted on by

在進行編譯視覺SLAM時，書中提到瞭ORB、SURF、SIFT提取方法，以及特征提取方法暴力匹配(Brute-Force Matcher)和快速近鄰匹配(FLANN)。以及7.9講述的3D-3D：迭代最近點(Iterative Closest Point,ICP)方法，ICP 的求解方式有兩種：利用線性代數求解(主要是SVD)，以及利用非線性優化方式求解。

完整代碼代碼如下：

鏈接：https://pan.baidu.com/s/1rlH9Jtg_aWtuYzmphqIJ3Q 提取碼：8888

main.cpp

#include <iostream>
 
#include "opencv2/opencv.hpp"
#include "opencv2/core/core.hpp"
#include "opencv2/features2d/features2d.hpp"
#include "opencv2/highgui/highgui.hpp"
#include <opencv2/xfeatures2d.hpp>
#include <iostream>
#include <vector>
#include <time.h>
#include <chrono>
#include <math.h>
#include<bits/stdc++.h>
 
 
using namespace std;
using namespace cv;
using namespace cv::xfeatures2d;
 
double picture1_size_change=1;
double picture2_size_change=1;
 
bool show_picture = true;
 
void extract_ORB2(string picture1, string picture2)
{
    //-- 讀取圖像
    Mat img_1 = imread(picture1, CV_LOAD_IMAGE_COLOR);
    Mat img_2 = imread(picture2, CV_LOAD_IMAGE_COLOR);
    assert(img_1.data != nullptr && img_2.data != nullptr);
    resize(img_1, img_1, Size(),  picture1_size_change, picture1_size_change);
    resize(img_2, img_2, Size(), picture2_size_change, picture2_size_change);
 
    //-- 初始化
    std::vector<KeyPoint> keypoints_1, keypoints_2;
    Mat descriptors_1, descriptors_2;
    Ptr<FeatureDetector> detector = ORB::create(2000,(1.200000048F), 8, 100);
    Ptr<DescriptorExtractor> descriptor = ORB::create(5000);
    Ptr<DescriptorMatcher> matcher = DescriptorMatcher::create("BruteForce-Hamming");
 
    //-- 第一步:檢測 Oriented FAST 角點位置
    chrono::steady_clock::time_point t1 = chrono::steady_clock::now();
    detector->detect(img_1, keypoints_1);
    detector->detect(img_2, keypoints_2);
 
    //-- 第二步:根據角點位置計算 BRIEF 描述子
    descriptor->compute(img_1, keypoints_1, descriptors_1);
    descriptor->compute(img_2, keypoints_2, descriptors_2);
    chrono::steady_clock::time_point t2 = chrono::steady_clock::now();
    chrono::duration<double> time_used = chrono::duration_cast<chrono::duration<double>>(t2 - t1);
    // cout << "extract ORB cost = " << time_used.count() * 1000 << " ms " << endl;
    cout << "detect " << keypoints_1.size() << " and " << keypoints_2.size() << " keypoints " << endl;
 
    if (show_picture)
    {
        Mat outimg1;
        drawKeypoints(img_1, keypoints_1, outimg1, Scalar::all(-1), DrawMatchesFlags::DEFAULT);
        imshow("ORB features", outimg1);
    }
 
    //-- 第三步:對兩幅圖像中的BRIEF描述子進行匹配，使用 Hamming 距離
    vector<DMatch> matches;
    // t1 = chrono::steady_clock::now();
    matcher->match(descriptors_1, descriptors_2, matches);
    t2 = chrono::steady_clock::now();
    time_used = chrono::duration_cast<chrono::duration<double>>(t2 - t1);
    cout << "extract and match ORB cost = " << time_used.count() * 1000 << " ms " << endl;
 
 
    //-- 第四步:匹配點對篩選
    // 計算最小距離和最大距離
    auto min_max = minmax_element(matches.begin(), matches.end(),
                                  [](const DMatch &m1, const DMatch &m2)
                                  { return m1.distance < m2.distance; });
    double min_dist = min_max.first->distance;
    double max_dist = min_max.second->distance;
 
    //   printf("-- Max dist : %f \n", max_dist);
    //   printf("-- Min dist : %f \n", min_dist);
 
    //當描述子之間的距離大於兩倍的最小距離時,即認為匹配有誤.但有時候最小距離會非常小,設置一個經驗值30作為下限.
    std::vector<DMatch> good_matches;
    for (int i = 0; i < descriptors_1.rows; i++)
    {
        if (matches[i].distance <= max(2 * min_dist, 30.0))
        {
            good_matches.push_back(matches[i]);
        }
    }
        cout << "match " << good_matches.size() << " keypoints " << endl;
 
 
    //-- 第五步:繪制匹配結果
    Mat img_match;
    Mat img_goodmatch;
    drawMatches(img_1, keypoints_1, img_2, keypoints_2, matches, img_match);
    drawMatches(img_1, keypoints_1, img_2, keypoints_2, good_matches, img_goodmatch);
 
    if (show_picture)
        imshow("good matches", img_goodmatch);
    if (show_picture)
        waitKey(0);
}
 
void extract_SIFT(string picture1, string picture2)
{
    // double t = (double)getTickCount();
    Mat temp = imread(picture1, IMREAD_GRAYSCALE);
    Mat image_check_changed = imread(picture2, IMREAD_GRAYSCALE);
    if (!temp.data || !image_check_changed.data)
    {
        printf("could not load images...\n");
        return;
    }
 
    resize(temp, temp, Size(),  picture1_size_change, picture1_size_change);
    resize(image_check_changed, image_check_changed, Size(), picture2_size_change, picture2_size_change);
 
    //Mat image_check_changed = Change_image(image_check);
    //("temp", temp);
    if (show_picture) 
        imshow("image_check_changed", image_check_changed);
 
    int minHessian = 500;
    // Ptr<SURF> detector = SURF::create(minHessian);    // surf
    Ptr<SIFT> detector = SIFT::create(); // sift
 
    vector<KeyPoint> keypoints_obj;
    vector<KeyPoint> keypoints_scene;
    Mat descriptor_obj, descriptor_scene;
 
 
    clock_t startTime, endTime;
    startTime = clock();
 
    chrono::steady_clock::time_point t1 = chrono::steady_clock::now();
    // cout << "extract ORB cost = " << time_used.count() * 1000 << " ms " << endl;
    detector->detectAndCompute(temp, Mat(), keypoints_obj, descriptor_obj);
    detector->detectAndCompute(image_check_changed, Mat(), keypoints_scene, descriptor_scene);
    cout << "detect " << keypoints_obj.size() << " and " << keypoints_scene.size() << " keypoints " << endl;
 
    // matching
    FlannBasedMatcher matcher;
    vector<DMatch> matches;
    matcher.match(descriptor_obj, descriptor_scene, matches);
 
    chrono::steady_clock::time_point t2 = chrono::steady_clock::now();
    chrono::duration<double> time_used = chrono::duration_cast<chrono::duration<double>>(t2 - t1);
    cout << "extract and match cost = " << time_used.count() * 1000 << " ms " << endl;
 
    //求最小最大距離
    double minDist = 1000;
    double maxDist = 0;
    //row--行 col--列
    for (int i = 0; i < descriptor_obj.rows; i++)
    {
        double dist = matches[i].distance;
        if (dist > maxDist)
        {
            maxDist = dist;
        }
        if (dist < minDist)
        {
            minDist = dist;
        }
    }
    // printf("max distance : %f\n", maxDist);
    // printf("min distance : %f\n", minDist);
 
    // find good matched points
    vector<DMatch> goodMatches;
    for (int i = 0; i < descriptor_obj.rows; i++)
    {
        double dist = matches[i].distance;
        if (dist < max(5 * minDist, 1.0))
        {
            goodMatches.push_back(matches[i]);
        }
    }
    //rectangle(temp, Point(1, 1), Point(177, 157), Scalar(0, 0, 255), 8, 0);
 
    cout << "match " << goodMatches.size() << " keypoints " << endl;
 
    endTime = clock();
    // cout << "took time : " << (double)(endTime - startTime) / CLOCKS_PER_SEC * 1000 << " ms" << endl;
 
    Mat matchesImg;
    drawMatches(temp, keypoints_obj, image_check_changed, keypoints_scene, goodMatches, matchesImg, Scalar::all(-1),
                Scalar::all(-1), vector<char>(), DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS);
    if (show_picture) 
        imshow("Flann Matching Result01", matchesImg);
    // imwrite("C:/Users/Administrator/Desktop/matchesImg04.jpg", matchesImg);
 
    //求h
    std::vector<Point2f> points1, points2;
 
    //保存對應點
    for (size_t i = 0; i < goodMatches.size(); i++)
    {
        //queryIdx是對齊圖像的描述子和特征點的下標。
        points1.push_back(keypoints_obj[goodMatches[i].queryIdx].pt);
        //queryIdx是是樣本圖像的描述子和特征點的下標。
        points2.push_back(keypoints_scene[goodMatches[i].trainIdx].pt);
    }
 
    // Find homography 計算Homography，RANSAC隨機抽樣一致性算法
    Mat H = findHomography(points1, points2, RANSAC);
    //imwrite("C:/Users/Administrator/Desktop/C-train/C-train/result/sift/Image4_SURF_MinHessian1000_ minDist1000_a0.9b70.jpg", matchesImg);
 
    vector<Point2f> obj_corners(4);
    vector<Point2f> scene_corners(4);
    obj_corners[0] = Point(0, 0);
    obj_corners[1] = Point(temp.cols, 0);
    obj_corners[2] = Point(temp.cols, temp.rows);
    obj_corners[3] = Point(0, temp.rows);
 
    //透視變換(把斜的圖片扶正)
    perspectiveTransform(obj_corners, scene_corners, H);
    //Mat dst;
    cvtColor(image_check_changed, image_check_changed, COLOR_GRAY2BGR);
    line(image_check_changed, scene_corners[0], scene_corners[1], Scalar(0, 0, 255), 2, 8, 0);
    line(image_check_changed, scene_corners[1], scene_corners[2], Scalar(0, 0, 255), 2, 8, 0);
    line(image_check_changed, scene_corners[2], scene_corners[3], Scalar(0, 0, 255), 2, 8, 0);
    line(image_check_changed, scene_corners[3], scene_corners[0], Scalar(0, 0, 255), 2, 8, 0);
 
 
    if (show_picture)
    {
        Mat outimg1;
        Mat temp_color = imread(picture1, CV_LOAD_IMAGE_COLOR);
        drawKeypoints(temp_color, keypoints_obj, outimg1, Scalar::all(-1), DrawMatchesFlags::DEFAULT);
        imshow("SIFT features", outimg1);
    }
 
    if (show_picture) 
        imshow("Draw object", image_check_changed);
    // imwrite("C:/Users/Administrator/Desktop/image04.jpg", image_check_changed);
 
    // t = ((double)getTickCount() - t) / getTickFrequency();
    // printf("averagetime:%f\n", t);
    if (show_picture) 
        waitKey(0);
}
 
void extract_SURF(string picture1, string picture2)
{
     // double t = (double)getTickCount();
    Mat temp = imread(picture1, IMREAD_GRAYSCALE);
    Mat image_check_changed = imread(picture2, IMREAD_GRAYSCALE);
    if (!temp.data || !image_check_changed.data)
    {
        printf("could not load images...\n");
        return;
    }
 
    resize(temp, temp, Size(),  picture1_size_change, picture1_size_change);
    resize(image_check_changed, image_check_changed, Size(), picture2_size_change, picture2_size_change);
 
    //Mat image_check_changed = Change_image(image_check);
    //("temp", temp);
    if (show_picture) 
        imshow("image_check_changed", image_check_changed);
 
    int minHessian = 500;
    Ptr<SURF> detector = SURF::create(minHessian);    // surf
    // Ptr<SIFT> detector = SIFT::create(minHessian); // sift
 
    vector<KeyPoint> keypoints_obj;
    vector<KeyPoint> keypoints_scene;
    Mat descriptor_obj, descriptor_scene;
 
 
    clock_t startTime, endTime;
    startTime = clock();
 
    chrono::steady_clock::time_point t1 = chrono::steady_clock::now();
    // cout << "extract ORB cost = " << time_used.count() * 1000 << " ms " << endl;
    detector->detectAndCompute(temp, Mat(), keypoints_obj, descriptor_obj);
    detector->detectAndCompute(image_check_changed, Mat(), keypoints_scene, descriptor_scene);
    cout << "detect " << keypoints_obj.size() << " and " << keypoints_scene.size() << " keypoints " << endl;
 
    // matching
    FlannBasedMatcher matcher;
    vector<DMatch> matches;
    matcher.match(descriptor_obj, descriptor_scene, matches);
 
    chrono::steady_clock::time_point t2 = chrono::steady_clock::now();
    chrono::duration<double> time_used = chrono::duration_cast<chrono::duration<double>>(t2 - t1);
    cout << "extract and match cost = " << time_used.count() * 1000 << " ms " << endl;
 
    //求最小最大距離
    double minDist = 1000;
    double maxDist = 0;
    //row--行 col--列
    for (int i = 0; i < descriptor_obj.rows; i++)
    {
        double dist = matches[i].distance;
        if (dist > maxDist)
        {
            maxDist = dist;
        }
        if (dist < minDist)
        {
            minDist = dist;
        }
    }
    // printf("max distance : %f\n", maxDist);
    // printf("min distance : %f\n", minDist);
 
    // find good matched points
    vector<DMatch> goodMatches;
    for (int i = 0; i < descriptor_obj.rows; i++)
    {
        double dist = matches[i].distance;
        if (dist < max(2 * minDist, 0.15))
        {
            goodMatches.push_back(matches[i]);
        }
    }
    //rectangle(temp, Point(1, 1), Point(177, 157), Scalar(0, 0, 255), 8, 0);
 
    cout << "match " << goodMatches.size() << " keypoints " << endl;
    endTime = clock();
    // cout << "took time : " << (double)(endTime - startTime) / CLOCKS_PER_SEC * 1000 << " ms" << endl;
 
    Mat matchesImg;
    drawMatches(temp, keypoints_obj, image_check_changed, keypoints_scene, goodMatches, matchesImg, Scalar::all(-1),
                Scalar::all(-1), vector<char>(), DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS);
    if (show_picture) 
        imshow("Flann Matching Result01", matchesImg);
    // imwrite("C:/Users/Administrator/Desktop/matchesImg04.jpg", matchesImg);
 
    //求h
    std::vector<Point2f> points1, points2;
 
    //保存對應點
    for (size_t i = 0; i < goodMatches.size(); i++)
    {
        //queryIdx是對齊圖像的描述子和特征點的下標。
        points1.push_back(keypoints_obj[goodMatches[i].queryIdx].pt);
        //queryIdx是是樣本圖像的描述子和特征點的下標。
        points2.push_back(keypoints_scene[goodMatches[i].trainIdx].pt);
    }
 
    // Find homography 計算Homography，RANSAC隨機抽樣一致性算法
    Mat H = findHomography(points1, points2, RANSAC);
    //imwrite("C:/Users/Administrator/Desktop/C-train/C-train/result/sift/Image4_SURF_MinHessian1000_ minDist1000_a0.9b70.jpg", matchesImg);
 
    vector<Point2f> obj_corners(4);
    vector<Point2f> scene_corners(4);
    obj_corners[0] = Point(0, 0);
    obj_corners[1] = Point(temp.cols, 0);
    obj_corners[2] = Point(temp.cols, temp.rows);
    obj_corners[3] = Point(0, temp.rows);
 
    //透視變換(把斜的圖片扶正)
    perspectiveTransform(obj_corners, scene_corners, H);
    //Mat dst;
    cvtColor(image_check_changed, image_check_changed, COLOR_GRAY2BGR);
    line(image_check_changed, scene_corners[0], scene_corners[1], Scalar(0, 0, 255), 2, 8, 0);
    line(image_check_changed, scene_corners[1], scene_corners[2], Scalar(0, 0, 255), 2, 8, 0);
    line(image_check_changed, scene_corners[2], scene_corners[3], Scalar(0, 0, 255), 2, 8, 0);
    line(image_check_changed, scene_corners[3], scene_corners[0], Scalar(0, 0, 255), 2, 8, 0);
 
 
    if (show_picture)
    {
        Mat outimg1;
        Mat temp_color = imread(picture1, CV_LOAD_IMAGE_COLOR);
        drawKeypoints(temp_color, keypoints_obj, outimg1, Scalar::all(-1), DrawMatchesFlags::DEFAULT);
        imshow("SURF features", outimg1);
    }
 
    if (show_picture) 
        imshow("Draw object", image_check_changed);
    // imwrite("C:/Users/Administrator/Desktop/image04.jpg", image_check_changed);
 
    // t = ((double)getTickCount() - t) / getTickFrequency();
    // printf("averagetime:%f\n", t);
    if (show_picture) 
        waitKey(0);
}
void extract_AKAZE(string picture1,string picture2)
{
    //讀取圖片
    Mat temp = imread(picture1,IMREAD_GRAYSCALE);
    Mat image_check_changed = imread(picture2,IMREAD_GRAYSCALE);
    //如果不能讀到其中任何一張圖片，則打印不能下載圖片
    if(!temp.data || !image_check_changed.data)
    {
        printf("could not load iamges...\n");
        return;
    }
    resize(temp,temp,Size(),picture1_size_change,picture1_size_change);
    resize(image_check_changed,image_check_changed,Size(),picture2_size_change,picture2_size_change);
 
    //Mat image_check_changed = Change_image(image_check);
    //("temp", temp);
 
    if(show_picture)
    {
        imshow("image_checked_changed",image_check_changed);
    }
    
    int minHessian=500;
    Ptr<AKAZE> detector=AKAZE::create();//AKAZE
 
    vector<keypoint> keypoints_obj;
    vector<keypoint> keypoints_scene;
    Mat descriptor_obj,descriptor_scene;
 
 
    clock_t startTime,endTime;
    startTime=clock();
 
    chrono::steady_clock::time_point t1=chrono::steady_clock::now();
    detector->detectAndCompute(temp,Mat(),keypoints_obj,descriptor_obj);
    detector->detectAndCompute(image_check_changed,Mat(),keypoints_scene,descriptor_scene);
    cout<<" detect "<<keypoints_obj.size()<<" and "<<keypoints_scene.size<<" keypoints "<<endl;
 
    //matching
    FlannBasedMatcher matcher;
    vector<DMatch> matches;
    matcher.match(descriptor_obj,descriptor_scene,matches);
 
    chrono::steady_clock::time_point t2 = chrono::steady_clock::now();
    chrono::duration<double> time_used = chrono::duration_cast<chrono::duration<double>>(t2-t1);
    cout << "extract and match cost = " << time_used.count()*1000<<" ms "<<endl;
 
    //求最小最大距離
    double minDist = 1000;
    double max_dist = 0;
    //row--行 col--列
    for(int i=0;i<descriptor_obj.rows;i++)
    {
        double dist = match[i].distance;
        if(dist > maxDist)
        {
            maxDist = dist;
        }
        if(dist<minDist)
        {
            minDist = dist;
        }
    }
    // printf("max distance : %f\n", maxDist);
    // printf("min distance : %f\n", minDist);
 
    // find good matched points
    vector<DMatch> goodMatches;
    for(imt i=0;i<descriptor_obj.rows;i++)
    {
        double dist = matches[i].distance;
        if(dist < max(5 * minDist,1.0))
        {
        goodMatches.push_back(matches[i]);
        }
    }
    //rectangle(temp, Point(1, 1), Point(177, 157), Scalar(0, 0, 255), 8, 0);
    cout<<" match "<<goodMatches.size()<<" keypoints "<<endl;
    endTime = clock();
    // cout << "took time : " << (double)(endTime - startTime) / CLOCKS_PER_SEC * 1000 << " ms" << endl;
    
    Mat matchesImg;
    drawMatches(temp,keypoints_obj,image_check_changed,keypoints_scene,goodMatches,
    matchesImg,Scalar::all(-1),
                   Scalar::all(-1),vector<char>(),DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS);
    if(show_picture)
       imshow("Flann Matching Result01",matchesImg);
    
    // imwrite("C:/Users/Administrator/Desktop/matchesImg04.jpg", matchesImg);
 
    //求h   
    std::vector<Point2f> points1,points2;
 
    //保存對應點
    for(size_t i = 0;i < goodMatches.size();i++)
    {
       //queryIdx是對齊圖像的描述子和特征點的下標。
        points1.push_back(keypoints_obj[goodMatches[i].queryIdx].pt);
        //queryIdx是是樣本圖像的描述子和特征點的下標。
        points2.push_back(keypoints_scene[goodMatches[i].trainIdx].pt);  
    }
    // Find homography 計算Homography，RANSAC隨機抽樣一致性算法
    Mat H = findHomography(points1,points2,RANSAC);
   //imwrite("C:/Users/Administrator/Desktop/C-train/C-train/result/sift/Image4_SURF_MinHessian1000_ minDist1000_a0.9b70.jpg", matchesImg);
 
   vector<Point2f> obj_corners(4);
   vector<Point2f> scene_corners(4);
   obj_corners[0] = Point(0,0);
   obj_corners[0] = Point(temp.count,0);
   obj_corners[0] = Point(temp.cols,temp.rows);
   obj_corners[0] = Point(0,temp.rows);
 
   //透視變換(把斜的圖片扶正)
   perspectiveTransform(obj_corners,scene_corners,H);
   //Mat dst
   cvtColor(image_check_changed,image_check_changed,COLOR_GRAY2BGR);
   line(image_check_changed,scene_corners[0],scene_corners[1],Scalar(0,0,255),2,8,0);
   line(image_check_changed,scene_corners[1],scene_corners[2],Scalar(0,0,255),2,8,0);
   line(image_check_changed,scene_corners[2],scene_corners[3],Scalar(0,0,255),2,8,0);
   line(image_check_changed,scene_corners[3],scene_corners[0],Scalar(0,0,255),2,8,0); 
 
   if(show_picture)
   {
       Mat outimg1;
       Mat temp_color = imread(picture1,CV_LOAD_IMAGE_COLOR);
       drawKeypoints(temp_color,keypoints_obj,outimg1,Scalar::all(-1),DrawMatchesFlags::DEFAULT);
       imshow("AKAZE features",outimg1);
   }
   if(show_picture)
      waitKey(0);
}
 
void extract_ORB(string picture1, string picture2)
{
    Mat img_1 = imread(picture1);
	Mat img_2 = imread(picture2);
 
    resize(img_1, img_1, Size(), picture1_size_change, picture1_size_change);
    resize(img_2, img_2, Size(), picture2_size_change, picture2_size_change);
 
	if (!img_1.data || !img_2.data)
	{
		cout << "error reading images " << endl;
		return ;
	}
 
	vector<Point2f> recognized;
	vector<Point2f> scene;
 
	recognized.resize(1000);
	scene.resize(1000);
 
	Mat d_srcL, d_srcR;
 
	Mat img_matches, des_L, des_R;
	//ORB算法的目標必須是灰度圖像
	cvtColor(img_1, d_srcL, COLOR_BGR2GRAY);//CPU版的ORB算法源碼中自帶對輸入圖像灰度化，此步可省略
	cvtColor(img_2, d_srcR, COLOR_BGR2GRAY);
 
	Ptr<ORB> d_orb = ORB::create(1500);
 
	Mat d_descriptorsL, d_descriptorsR, d_descriptorsL_32F, d_descriptorsR_32F;
 
	vector<KeyPoint> keyPoints_1, keyPoints_2;
 
	//設置關鍵點間的匹配方式為NORM_L2，更建議使用 FLANNBASED = 1, BRUTEFORCE = 2, BRUTEFORCE_L1 = 3, BRUTEFORCE_HAMMING = 4, BRUTEFORCE_HAMMINGLUT = 5, BRUTEFORCE_SL2 = 6 
	Ptr<DescriptorMatcher> d_matcher = DescriptorMatcher::create(NORM_L2);
 
	std::vector<DMatch> matches;//普通匹配
	std::vector<DMatch> good_matches;//通過keyPoint之間距離篩選匹配度高的匹配結果
 
    clock_t startTime, endTime;
    startTime = clock();
 
 
 
    chrono::steady_clock::time_point t1 = chrono::steady_clock::now();
	d_orb -> detectAndCompute(d_srcL, Mat(), keyPoints_1, d_descriptorsL);
	d_orb -> detectAndCompute(d_srcR, Mat(), keyPoints_2, d_descriptorsR);
    cout << "detect " << keyPoints_1.size() << " and " << keyPoints_2.size() << " keypoints " << endl;
    // endTime = clock();
    // cout << "took time : " << (double)(endTime - startTime) / CLOCKS_PER_SEC * 1000 << " ms" << endl;
 
 
	d_matcher -> match(d_descriptorsL, d_descriptorsR, matches);//L、R表示左右兩幅圖像進行匹配
 
    //計算匹配所需時間
    chrono::steady_clock::time_point t2 = chrono::steady_clock::now();
    chrono::duration<double> time_used = chrono::duration_cast<chrono::duration<double>>(t2 - t1);
    cout << "extract and match cost = " << time_used.count() * 1000 << " ms " << endl;
 
 
	int sz = matches.size();
	double max_dist = 0; double min_dist = 100;
 
	for (int i = 0; i < sz; i++)
	{
		double dist = matches[i].distance;
		if (dist < min_dist) min_dist = dist;
		if (dist > max_dist) max_dist = dist;
	}
  
	for (int i = 0; i < sz; i++)
	{
		if (matches[i].distance < 0.6*max_dist)
		{
			good_matches.push_back(matches[i]);
		}
	}
 
    cout << "match " << good_matches.size() << " keypoints " << endl;
    // endTime = clock();
    // cout << "took time : " << (double)(endTime - startTime) / CLOCKS_PER_SEC * 1000 << " ms" << endl;
 
 
	//提取良好匹配結果中在待測圖片上的點集，確定匹配的大概位置
	for (size_t i = 0; i < good_matches.size(); ++i)
	{
		scene.push_back(keyPoints_2[ good_matches[i].trainIdx ].pt);
	}
 
	for(unsigned int j = 0; j < scene.size(); j++)
		cv::circle(img_2, scene[j], 2, cv::Scalar(0, 255, 0), 2);
	//畫出普通匹配結果
	Mat ShowMatches;
	drawMatches(img_1,keyPoints_1,img_2,keyPoints_2,matches,ShowMatches);
	if (show_picture) 
        imshow("matches", ShowMatches);
	// imwrite("matches.png", ShowMatches);
	//畫出良好匹配結果
	Mat ShowGoodMatches;
	drawMatches(img_1,keyPoints_1,img_2,keyPoints_2,good_matches,ShowGoodMatches);
	if (show_picture) 
        imshow("good_matches", ShowGoodMatches);
	// imwrite("good_matches.png", ShowGoodMatches);
	//畫出良好匹配結果中在待測圖片上的點集
	if (show_picture) 
        imshow("MatchPoints_in_img_2", img_2);
	// imwrite("MatchPoints_in_img_2.png", img_2);
	if (show_picture) 
        waitKey(0);
}
 
int main(int argc, char **argv)
{
    string picture1=string(argv[1]);
    string picture2=string(argv[2]);
    // string picture1 = "data/picture1/6.jpg";
    // string picture2 = "data/picture2/16.PNG";
 
    cout << "\nextract_ORB::" << endl;
    extract_ORB(picture1, picture2);
 
    cout << "\nextract_ORB::" << endl;
    extract_ORB2(picture1, picture2);
 
    cout << "\nextract_SURF::" << endl;
    extract_SURF(picture1, picture2);
 
      cout << "\nextract_AKAZE::" << endl;
    extract_AKAZE(picture1, picture2);
   
    cout << "\nextract_SIFT::" << endl;
    extract_SIFT(picture1, picture2);
    cout << "success!!" << endl;
}

CMakeLists.txt

CMAKE_MINIMUM_REQUIRED(VERSION 2.8.3)  # 設定版本
PROJECT(DescriptorCompare) # 設定工程名
SET(CMAKE_CXX_COMPILER "g++")  # 設定編譯器
add_compile_options(-std=c++14)   #編譯選項，選擇c++版本
 
# 設定可執行二進制文件的目錄（最後生成的可執行文件放置的目錄）
SET(EXECUTABLE_OUTPUT_PATH ${PROJECT_SOURCE_DIR})
 
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -fpermissive -g -O3  -Wno-unused-function  -Wno-return-type")
 
find_package(OpenCV 3.0 REQUIRED)
 
 
message(STATUS "Using opencv version ${OpenCV_VERSION}")
find_package(Eigen3 3.3.8 REQUIRED)
find_package(Pangolin REQUIRED)
 
 
 
 
 
# 設定鏈接目錄
LINK_DIRECTORIES(${PROJECT_SOURCE_DIR}/lib)
 
# 設定頭文件目錄
INCLUDE_DIRECTORIES(
    ${PROJECT_SOURCE_DIR}/include
    ${EIGEN3_INCLUDE_DIR}
    ${OpenCV_INCLUDE_DIR}
    ${Pangolin_INCLUDE_DIRS}
    )
    
 
add_library(${PROJECT_NAME}
test.cc
)
 
target_link_libraries( ${PROJECT_NAME}
${OpenCV_LIBS}
${EIGEN3_LIBS}
${Pangolin_LIBRARIES}
 
)
 
 
add_executable(main main.cpp )
target_link_libraries(main ${PROJECT_NAME} )
 
 
add_executable(icp icp.cpp )
target_link_libraries(icp ${PROJECT_NAME} )

執行效果

./main 1.png 2.png 
extract_ORB::
detect 1500 and 1500 keypoints 
extract and match cost = 21.5506 ms 
match 903 keypoints 
 
extract_ORB::
detect 1304 and 1301 keypoints 
extract and match ORB cost = 25.4976 ms 
match 313 keypoints 
 
extract_SURF::
detect 915 and 940 keypoints 
extract and match cost = 53.8371 ms 
match 255 keypoints 
 
extract_SIFT::
detect 1536 and 1433 keypoints 
extract and match cost = 97.9322 ms 
match 213 keypoints 
success!!

ICP

#include <iostream>
#include <opencv2/core/core.hpp>
#include <opencv2/features2d/features2d.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/calib3d/calib3d.hpp>
#include <Eigen/Core>
#include <Eigen/Dense>
#include <Eigen/Geometry>
#include <Eigen/SVD>
 
#include <pangolin/pangolin.h>
#include <chrono>
 
 
using namespace std;
using namespace cv;
 
int picture_h=480;
int picture_w=640;
 
 
bool show_picture = true;
 
 
void find_feature_matches(
  const Mat &img_1, const Mat &img_2,
  std::vector<KeyPoint> &keypoints_1,
  std::vector<KeyPoint> &keypoints_2,
  std::vector<DMatch> &matches);
 
// 像素坐標轉相機歸一化坐標
Point2d pixel2cam(const Point2d &p, const Mat &K);
 
void pose_estimation_3d3d(
  const vector<Point3f> &pts1,
  const vector<Point3f> &pts2,
  Mat &R, Mat &t
);
 
 
 
int main(int argc, char **argv) {
  if (argc != 5) {
    cout << "usage: pose_estimation_3d3d img1 img2 depth1 depth2" << endl;
    return 1;
  }
  //-- 讀取圖像
  Mat img_1 = imread(argv[1], CV_LOAD_IMAGE_COLOR);
  Mat img_2 = imread(argv[2], CV_LOAD_IMAGE_COLOR);
 
  vector<KeyPoint> keypoints_1, keypoints_2;
  vector<DMatch> matches;
  find_feature_matches(img_1, img_2, keypoints_1, keypoints_2, matches);
  cout << "picture1 keypoints: " << keypoints_1.size() << " \npicture2 keypoints: " << keypoints_2.size() << endl;
  cout << "一共找到瞭 " << matches.size() << " 組匹配點" << endl;
 
  // 建立3D點
  Mat depth1 = imread(argv[3], CV_8UC1);       // 深度圖為16位無符號數，單通道圖像
  Mat depth2 = imread(argv[4], CV_8UC1);       // 深度圖為16位無符號數，單通道圖像
  Mat K = (Mat_<double>(3, 3) << 595.2, 0, 328.9, 0, 599.0, 253.9, 0, 0, 1);
  vector<Point3f> pts1, pts2;
 
  for (DMatch m:matches) {
    int d1 = 255-(int)depth1.ptr<uchar>(int(keypoints_1[m.queryIdx].pt.y))[int(keypoints_1[m.queryIdx].pt.x)];
    int d2 = 255-(int)depth2.ptr<uchar>(int(keypoints_2[m.trainIdx].pt.y))[int(keypoints_2[m.trainIdx].pt.x)];
    if (d1 == 0 || d2 == 0)   // bad depth
      continue;
    Point2d p1 = pixel2cam(keypoints_1[m.queryIdx].pt, K);
    Point2d p2 = pixel2cam(keypoints_2[m.trainIdx].pt, K);
    float dd1 = int(d1) / 1000.0;
    float dd2 = int(d2) / 1000.0;
    pts1.push_back(Point3f(p1.x * dd1, p1.y * dd1, dd1));
    pts2.push_back(Point3f(p2.x * dd2, p2.y * dd2, dd2));
  }
 
  cout << "3d-3d pairs: " << pts1.size() << endl;
  Mat R, t;
  pose_estimation_3d3d(pts1, pts2, R, t);
 
 
 
 
 
//DZQ ADD
  cv::Mat Pose = (Mat_<double>(4, 4) << R.at<double>(0, 0), R.at<double>(0, 1), R.at<double>(0, 2), t.at<double>(0),
                  R.at<double>(1, 0), R.at<double>(1, 1), R.at<double>(1, 2), t.at<double>(1),
                  R.at<double>(2, 0), R.at<double>(2, 1), R.at<double>(2, 2), t.at<double>(2),
                  0, 0, 0, 1);
 
  cout << "[delete outliers] Matched objects distance: ";
  vector<double> vDistance;
  double allDistance = 0; //存儲總距離，用來求平均匹配距離，用平均的誤差距離來剔除外點
  for (int i = 0; i < pts1.size(); i++)
  {
    Mat point = Pose * (Mat_<double>(4, 1) << pts2[i].x, pts2[i].y, pts2[i].z, 1);
    double distance = pow(pow(pts1[i].x - point.at<double>(0), 2) + pow(pts1[i].y - point.at<double>(1), 2) + pow(pts1[i].z - point.at<double>(2), 2), 0.5);
    vDistance.push_back(distance);
    allDistance += distance;
    // cout << distance << " ";
  }
  // cout << endl;
  double avgDistance = allDistance / pts1.size(); //求一個平均距離
  int N_outliers = 0;
  for (int i = 0, j = 0; i < pts1.size(); i++, j++) //i用來記錄剔除後vector遍歷的位置，j用來記錄原位置
  {
    if (vDistance[i] > 1.5 * avgDistance) //匹配物體超過平均距離的N倍就會被剔除 [delete outliers]  DZQ FIXED_PARAM
    {
      N_outliers++;
    }
  }
  cout << "N_outliers:: " << N_outliers << endl;
 
 
 
 
 
 
  // show points
  {
    //創建一個窗口
    pangolin::CreateWindowAndBind("show points", 640, 480);
    //啟動深度測試
    glEnable(GL_DEPTH_TEST);
 
    // Define Projection and initial ModelView matrix
    pangolin::OpenGlRenderState s_cam(
        pangolin::ProjectionMatrix(640, 480, 420, 420, 320, 240, 0.05, 500),
        //對應的是gluLookAt,攝像機位置,參考點位置,up vector(上向量)
        pangolin::ModelViewLookAt(0, -5, 0.1, 0, 0, 0, pangolin::AxisY));
 
    // Create Interactive View in window
    pangolin::Handler3D handler(s_cam);
    //setBounds 跟opengl的viewport 有關
    //看SimpleDisplay中邊界的設置就知道
    pangolin::View &d_cam = pangolin::CreateDisplay()
                                .SetBounds(0.0, 1.0, 0.0, 1.0, -640.0f / 480.0f)
                                .SetHandler(&handler);
 
    while (!pangolin::ShouldQuit())
    {
 
      // Clear screen and activate view to render into
      glClearColor(0.97,0.97,1.0, 1); //背景色
 
      glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
      d_cam.Activate(s_cam);
 
 
      glBegin(GL_POINTS);  //繪制匹配點
      glLineWidth(5);
      for (int i = 0; i < pts1.size(); i++)
      {
        glColor3f(1, 0, 0);
        glVertex3d(pts1[i].x,pts1[i].y,pts1[i].z);
        Mat point = Pose * (Mat_<double>(4, 1) << pts2[i].x, pts2[i].y, pts2[i].z, 1);
        glColor3f(0, 1, 0);
        glVertex3d(point.at<double>(0),point.at<double>(1),point.at<double>(2));
      }
      glEnd();
 
 
      glBegin(GL_LINES);    //繪制匹配線
      glLineWidth(1);
      for (int i = 0; i < pts1.size(); i++)
      {
        glColor3f(0, 0, 1);
        glVertex3d(pts1[i].x,pts1[i].y,pts1[i].z);
        Mat point = Pose * (Mat_<double>(4, 1) << pts2[i].x, pts2[i].y, pts2[i].z, 1);
        glVertex3d(point.at<double>(0),point.at<double>(1),point.at<double>(2));
      }
      glEnd();
 
 
 
 
      glBegin(GL_POINTS);    //繪制所有點
      glLineWidth(5);
      glColor3f(1, 0.5, 0);
      for (int i = 0; i < picture_h; i+=2)
      {
        for (int j = 0; j < picture_w; j+=2)
        {
          int d1 = 255-(int)depth1.ptr<uchar>(i)[j];
          if (d1 == 0) // bad depth
            continue;
          Point2d temp_p;
          temp_p.y=i;    //這裡的x和y應該和i j相反
          temp_p.x=j;
          Point2d p1 = pixel2cam(temp_p, K);
          float dd1 = int(d1) / 1000.0;
          glVertex3d(p1.x * dd1, p1.y * dd1, dd1);
          // glVertex3d(j/1000.0, i/1000.0, d1/200.0);
        }
      }
      glEnd();
 
 
 
      // Swap frames and Process Events
      pangolin::FinishFrame();
    }
  }
}
 
void find_feature_matches(const Mat &img_1, const Mat &img_2,
                          std::vector<KeyPoint> &keypoints_1,
                          std::vector<KeyPoint> &keypoints_2,
                          std::vector<DMatch> &matches) {
  //-- 初始化
  Mat descriptors_1, descriptors_2;
  // used in OpenCV3
  Ptr<FeatureDetector> detector = ORB::create(2000,(1.200000048F), 8, 100);
  Ptr<DescriptorExtractor> descriptor = ORB::create(5000);
  
  Ptr<DescriptorMatcher> matcher = DescriptorMatcher::create("BruteForce-Hamming");
  //-- 第一步:檢測 Oriented FAST 角點位置
  detector->detect(img_1, keypoints_1);
  detector->detect(img_2, keypoints_2);
 
  //-- 第二步:根據角點位置計算 BRIEF 描述子
  descriptor->compute(img_1, keypoints_1, descriptors_1);
  descriptor->compute(img_2, keypoints_2, descriptors_2);
 
  //-- 第三步:對兩幅圖像中的BRIEF描述子進行匹配，使用 Hamming 距離
  vector<DMatch> match;
  // BFMatcher matcher ( NORM_HAMMING );
  matcher->match(descriptors_1, descriptors_2, match);
 
  //-- 第四步:匹配點對篩選
  double min_dist = 10000, max_dist = 0;
 
  //找出所有匹配之間的最小距離和最大距離, 即是最相似的和最不相似的兩組點之間的距離
  for (int i = 0; i < descriptors_1.rows; i++) {
    double dist = match[i].distance;
    if (dist < min_dist) min_dist = dist;
    if (dist > max_dist) max_dist = dist;
  }
 
  printf("-- Max dist : %f \n", max_dist);
  printf("-- Min dist : %f \n", min_dist);
 
  //當描述子之間的距離大於兩倍的最小距離時,即認為匹配有誤.但有時候最小距離會非常小,設置一個經驗值30作為下限.
  for (int i = 0; i < descriptors_1.rows; i++) {
    if (match[i].distance <= max(2 * min_dist, 30.0)) {
      matches.push_back(match[i]);
    }
  }
 
  //-- 第五步:繪制匹配結果
  if(show_picture)
  {
 
    Mat img_match;
    Mat img_goodmatch;
    drawMatches(img_1, keypoints_1, img_2, keypoints_2, matches, img_match);
    imshow("all matches", img_match);
    waitKey(0);
  }
}
 
Point2d pixel2cam(const Point2d &p, const Mat &K) {
  return Point2d(
    (p.x - K.at<double>(0, 2)) / K.at<double>(0, 0),
    (p.y - K.at<double>(1, 2)) / K.at<double>(1, 1)
  );
}
 
void pose_estimation_3d3d(const vector<Point3f> &pts1,
                          const vector<Point3f> &pts2,
                          Mat &R, Mat &t) {
  Point3f p1, p2;     // center of mass
  int N = pts1.size();
  for (int i = 0; i < N; i++) {
    p1 += pts1[i];
    p2 += pts2[i];
  }
  p1 = Point3f(Vec3f(p1) / N);
  p2 = Point3f(Vec3f(p2) / N);
  vector<Point3f> q1(N), q2(N); // remove the center
  for (int i = 0; i < N; i++) {
    q1[i] = pts1[i] - p1;
    q2[i] = pts2[i] - p2;
  }
 
  // compute q1*q2^T
  Eigen::Matrix3d W = Eigen::Matrix3d::Zero();
  for (int i = 0; i < N; i++) {
    W += Eigen::Vector3d(q1[i].x, q1[i].y, q1[i].z) * Eigen::Vector3d(q2[i].x, q2[i].y, q2[i].z).transpose();
  }
  // cout << "W=" << W << endl;
 
  // SVD on W
  Eigen::JacobiSVD<Eigen::Matrix3d> svd(W, Eigen::ComputeFullU | Eigen::ComputeFullV);
  Eigen::Matrix3d U = svd.matrixU();
  Eigen::Matrix3d V = svd.matrixV();
 
 
  Eigen::Matrix3d R_ = U * (V.transpose());
  if (R_.determinant() < 0) {
    R_ = -R_;
  }
  Eigen::Vector3d t_ = Eigen::Vector3d(p1.x, p1.y, p1.z) - R_ * Eigen::Vector3d(p2.x, p2.y, p2.z);
 
  // convert to cv::Mat
  R = (Mat_<double>(3, 3) <<
    R_(0, 0), R_(0, 1), R_(0, 2),
    R_(1, 0), R_(1, 1), R_(1, 2),
    R_(2, 0), R_(2, 1), R_(2, 2)
  );
  t = (Mat_<double>(3, 1) << t_(0, 0), t_(1, 0), t_(2, 0));
}
 
 
 
 
 
void convertRGB2Gray(string picture)
{
 
 
	double min;
	double max;
  Mat depth_new_1 = imread(picture);       // 深度圖為16位無符號數，單通道圖像
	
Mat test=Mat(20,256,CV_8UC3);
	int s;
	for (int i = 0; i < 20; i++) {
  std::cout<<i<<" ";
 
		Vec3b* p = test.ptr<Vec3b>(i);
		for (s = 0; s < 32; s++) {
			p[s][0] = 128 + 4 * s;
			p[s][1] = 0;
			p[s][2] = 0;
		}
		p[32][0] = 255;
		p[32][1] = 0;
		p[32][2] = 0;
		for (s = 0; s < 63; s++) {
			p[33+s][0] = 255;
			p[33+s][1] = 4+4*s;
			p[33+s][2] = 0;
		}
		p[96][0] = 254;
		p[96][1] = 255;
		p[96][2] = 2;
		for (s = 0; s < 62; s++) {
			p[97 + s][0] = 250 - 4 * s;
			p[97 + s][1] = 255;
			p[97 + s][2] = 6+4*s;
		}
		p[159][0] = 1;
		p[159][1] = 255;
		p[159][2] = 254;
		for (s = 0; s < 64; s++) {
			p[160 + s][0] = 0;
			p[160 + s][1] = 252 - (s * 4);
			p[160 + s][2] = 255;
		}
		for (s = 0; s < 32; s++) {
			p[224 + s][0] = 0;
			p[224 + s][1] = 0;
			p[224 + s][2] = 252-4*s;
		}
	}
 
  cout<<"depth_new_1 :: "<<depth_new_1.cols<<" "<<depth_new_1.rows<<" "<<endl;
 
 
Mat img_g=Mat(picture_h,picture_w,CV_8UC1);
for(int i=0;i<picture_h;i++)
{
 
  Vec3b *p = test.ptr<Vec3b>(0);
  Vec3b *q = depth_new_1.ptr<Vec3b>(i);
  for (int j = 0; j < picture_w; j++)
  {
    
    for(int k=0;k<256;k++)
    {
      if ( (((int)p[k][0] - (int)q[j][0] < 4) && ((int)q[j][0] - (int)p[k][0] < 4))&&
           (((int)p[k][1] - (int)q[j][1] < 4) && ((int)q[j][1] - (int)p[k][1] < 4))&&
           (((int)p[k][2] - (int)q[j][2] < 4) && ((int)q[j][2] - (int)p[k][2] < 4)))
     
      {
        img_g.at<uchar>(i,j)=k;
      }
    }
  }
}
 
	imwrite("14_Depth_3.png", img_g);
	waitKey();
 
 
}

CMakeLists.txt

和上面一樣。

./icp 1.png 2.png 1_depth.png 2_depth.png 

-- Max dist : 87.000000 
-- Min dist : 4.000000 
picture1 keypoints: 1304 
picture2 keypoints: 1301
一共找到瞭 313 組匹配點
3d-3d pairs: 313
[delete outliers] Matched objects distance: N_outliers:: 23

執行效果

以上就是淺析ORB、SURF、SIFT特征點提取方法以及ICP匹配方法的詳細內容，更多關於特征點提取方法 ICP匹配方法的資料請關註WalkonNet其它相關文章！

推薦閱讀：