OpenCV3.0实现SIFT特征提取+RANSAC剔除误匹配点

蔚落 2022-05-22 01:44 750阅读 0赞

开发环境：VS2013+OpenCV3.0

一、Preparation

在学习图像识别中特征点检测与匹配时，需要用到OpenCV中的SIFT和SURF算法，如SiftFeatureDetector或SiftFeatureExtractor,在OpenCV2中SIFT和SURF算法被写在文件\#include<opencv2/nonfree/feature2d.hpp>中，但是OpenCV3.0中这两种算法存放在opencv\_contrib目录下，若想使用该目录下的功能需要自己重新编译OpenCV。

SIFT算法详解：[SIFT算法详解][SIFT]

SIFT算法译文：[SIFT算法译文][SIFT 1]

下载opencv和opencv\_contrib: [https://github.com/opencv][https_github.com_opencv]

opencv3.0+opencv\_contrib在Windows上的配置和编译可参考：

[OpenCV学习笔记（八）—— OpenCV 3.1.0 + opencv\_contrib编译（Windows）\_linshuhe1的专栏-CSDN博客\_opencv编译时highgui310][OpenCV_ OpenCV 3.1.0 _ opencv_contrib_Windows_linshuhe1_-CSDN_opencv_highgui310]

Note：这里opencv和opencv\_contrib的版本要一致，否则后面编译的时候会报错。

二、RANSAC Algorithm

RANSAC（RANdom SAmple Consensus), 随机抽样一致性，它可以从一组包含“局外点”的观测数据，通过迭代的方式训练最优的参数模型，不符合最优参数模型的被定义为“局外点”。

1. RANSAC消除误匹配点可以分为三部分：  
（1）根据matches将特征点对齐，将坐标转换为float类型

（2）使用求基础矩阵的方法，findFundamentalMat，得到RansacStatus

（3）根据RansacStatus来删除误匹配点，即RansacStatus\[0\]=0的点。

2.函数说明

findFundamentalMat()得到基础矩阵，基础矩阵对应三维图像中像点之间的对应关系。

findHomography()得到变换矩阵。

2. 优点与缺点

优点是对模型参数的估计具有鲁棒性，能从具有大量局外点的观测数据中估计出高精度的模型参数，缺点就是迭代次数没有上限，如果设置迭代次数很可能得不到最优的参数模型，甚至得到错误的模型。RANSAC只有一定的概率得到可信模型，迭代次数越大，概率越高，两者成正比。

RANSAC只能从数据集中得到一个模型，如果有多个模型，该算法不能实现。

三. 代码实现与效果

#include <iostream>
    #include<opencv2/highgui/highgui.hpp>
    #include<opencv2/imgproc/imgproc.hpp>
    #include <opencv2/opencv.hpp> 
    #include<opencv2/xfeatures2d.hpp>
    #include<opencv2/core/core.hpp>
    
    using namespace cv;  
    using namespace std;
    using namespace cv::xfeatures2d;//只有加上这句命名空间，SiftFeatureDetector and SiftFeatureExtractor才可以使用
    
    int main()
    {
    	//Create SIFT class pointer
    	Ptr<Feature2D> f2d = xfeatures2d::SIFT::create();
    	//SiftFeatureDetector siftDetector;
    	//Loading images
    	Mat img_1 = imread("101200.jpg");
    	Mat img_2 = imread("101201.jpg");
    	if (!img_1.data || !img_2.data)
    	{
    		cout << "Reading picture error！" << endl;
    		return false;
    	}
    	//Detect the keypoints
    	double t0 = getTickCount();//当前滴答数
    	vector<KeyPoint> keypoints_1, keypoints_2;
    	f2d->detect(img_1, keypoints_1);
    	f2d->detect(img_2, keypoints_2);
    	cout << "The keypoints number of img1 is:" << keypoints_1.size() << endl;
    	cout << "The keypoints number of img2 is:" << keypoints_2.size() << endl;
    	//Calculate descriptors (feature vectors)
    	Mat descriptors_1, descriptors_2;
    	f2d->compute(img_1, keypoints_1, descriptors_1);
    	f2d->compute(img_2, keypoints_2, descriptors_2);
    	double freq = getTickFrequency();
    	double tt = ((double)getTickCount() - t0) / freq;
    	cout << "Extract SIFT Time:" <<tt<<"ms"<< endl;
    	//画关键点
    	Mat img_keypoints_1, img_keypoints_2;
    	drawKeypoints(img_1,keypoints_1,img_keypoints_1,Scalar::all(-1),0);
    	drawKeypoints(img_2, keypoints_2, img_keypoints_2, Scalar::all(-1), 0);
    	//imshow("img_keypoints_1",img_keypoints_1);
    	//imshow("img_keypoints_2",img_keypoints_2);
    
    	//Matching descriptor vector using BFMatcher
    	BFMatcher matcher;
    	vector<DMatch> matches;
    	matcher.match(descriptors_1, descriptors_2, matches);
    	cout << "The number of match:" << matches.size()<<endl;
    	//绘制匹配出的关键点
    	Mat img_matches;
    	drawMatches(img_1, keypoints_1, img_2, keypoints_2, matches, img_matches);
    	//imshow("Match image",img_matches);
    	//计算匹配结果中距离最大和距离最小值
    	double min_dist = matches[0].distance, max_dist = matches[0].distance;
    	for (int m = 0; m < matches.size(); m++)
    	{
    		if (matches[m].distance<min_dist)
    		{
    			min_dist = matches[m].distance;
    		}
    		if (matches[m].distance>max_dist)
    		{
    			max_dist = matches[m].distance;
    		}	
    	}
    	cout << "min dist=" << min_dist << endl;
    	cout << "max dist=" << max_dist << endl;
    	//筛选出较好的匹配点
    	vector<DMatch> goodMatches;
    	for (int m = 0; m < matches.size(); m++)
    	{
    		if (matches[m].distance < 0.6*max_dist)
    		{
    			goodMatches.push_back(matches[m]);
    		}
    	}
    	cout << "The number of good matches:" <<goodMatches.size()<< endl;
    	//画出匹配结果
    	Mat img_out;
    	//红色连接的是匹配的特征点数，绿色连接的是未匹配的特征点数
    	//matchColor – Color of matches (lines and connected keypoints). If matchColor==Scalar::all(-1) , the color is generated randomly.
    	//singlePointColor – Color of single keypoints(circles), which means that keypoints do not have the matches.If singlePointColor == Scalar::all(-1), the color is generated randomly.
    	//CV_RGB(0, 255, 0)存储顺序为R-G-B,表示绿色
    	drawMatches(img_1, keypoints_1, img_2, keypoints_2, goodMatches, img_out, Scalar::all(-1), CV_RGB(0, 0, 255), Mat(), 2);
    	imshow("good Matches",img_out);
        //RANSAC匹配过程
    	vector<DMatch> m_Matches;
    	m_Matches = goodMatches;
    	int ptCount = goodMatches.size();
    	if (ptCount < 100)
    	{
    		cout << "Don't find enough match points" << endl;
    		return 0;
    	}
    
    	//坐标转换为float类型
    	vector <KeyPoint> RAN_KP1, RAN_KP2;
    	//size_t是标准C库中定义的，应为unsigned int，在64位系统中为long unsigned int,在C++中为了适应不同的平台，增加可移植性。
    	for (size_t i = 0; i < m_Matches.size(); i++)
    	{
    		RAN_KP1.push_back(keypoints_1[goodMatches[i].queryIdx]);
    		RAN_KP2.push_back(keypoints_2[goodMatches[i].trainIdx]);
    		//RAN_KP1是要存储img01中能与img02匹配的点
    		//goodMatches存储了这些匹配点对的img01和img02的索引值
    	}
    	//坐标变换
    	vector <Point2f> p01, p02;
    	for (size_t i = 0; i < m_Matches.size(); i++)
    	{
    		p01.push_back(RAN_KP1[i].pt);
    		p02.push_back(RAN_KP2[i].pt);
    	}
    	/*vector <Point2f> img1_corners(4);
    	img1_corners[0] = Point(0,0);
    	img1_corners[1] = Point(img_1.cols,0);
    	img1_corners[2] = Point(img_1.cols, img_1.rows);
    	img1_corners[3] = Point(0, img_1.rows);
    	vector <Point2f> img2_corners(4);*/
    	求转换矩阵
    	//Mat m_homography;
    	//vector<uchar> m;
    	//m_homography = findHomography(p01, p02, RANSAC);//寻找匹配图像
    	//求基础矩阵 Fundamental,3*3的基础矩阵
    	vector<uchar> RansacStatus;
    	Mat Fundamental = findFundamentalMat(p01, p02, RansacStatus, FM_RANSAC);
    	//重新定义关键点RR_KP和RR_matches来存储新的关键点和基础矩阵，通过RansacStatus来删除误匹配点
    	vector <KeyPoint> RR_KP1, RR_KP2;
    	vector <DMatch> RR_matches;
    	int index = 0;
    	for (size_t i = 0; i < m_Matches.size(); i++)
    	{
    		if (RansacStatus[i] != 0)
    		{
    			RR_KP1.push_back(RAN_KP1[i]);
    			RR_KP2.push_back(RAN_KP2[i]);
    			m_Matches[i].queryIdx = index;
    			m_Matches[i].trainIdx = index;
    			RR_matches.push_back(m_Matches[i]);
    			index++;
    		}
    	}
    	cout << "RANSAC后匹配点数" <<RR_matches.size()<< endl;
    	Mat img_RR_matches;
    	drawMatches(img_1, RR_KP1, img_2, RR_KP2, RR_matches, img_RR_matches);
    	imshow("After RANSAC",img_RR_matches);
    	//等待任意按键按下
    	waitKey(0);
    }

运行效果：

![20180612173113100][]

![20180612173123869][]

完整的工程文件链接：[OpenCV3.0实现SIFT特征提取+RANSAC剔除误匹配点\_opencv3.0特征匹配,opencvransac特征匹配-C++文档类资源-CSDN下载][OpenCV3.0_SIFT_RANSAC_opencv3.0_opencvransac_-C_-CSDN]

[SIFT]: https://blog.csdn.net/zddblog/article/details/7521424
[SIFT 1]: http://www.cnblogs.com/cuteshongshong/archive/2012/05/25/2506374.html
[https_github.com_opencv]: https://github.com/opencv
[OpenCV_ OpenCV 3.1.0 _ opencv_contrib_Windows_linshuhe1_-CSDN_opencv_highgui310]: https://blog.csdn.net/linshuhe1/article/details/51221015
[20180612173113100]: /images/20220522/ec5ae7622d2c4701bb3f8d87e1129602.png
[20180612173123869]: /images/20220522/c9506fb08ed3454dad49a454b2fe6b4b.png
[OpenCV3.0_SIFT_RANSAC_opencv3.0_opencvransac_-C_-CSDN]: https://download.csdn.net/download/lindamtd/10798831