利用OpenCV(2.4.13)的SURF/SIFT + RANSAC + 加权融合实现图像拼接(vs17)

// ImageMosaic.cpp : 此文件包含 "main" 函数。程序执行将在此处开始并结束。 // source code : https://blog.csdn.net/dcrmg/article/details/52629856 // explanation : https://blog.csdn.net/u012384044/article/details/73162675#commentBox // #include "pch.h" #include <iostream> #include <string.h> #include "opencv2/highgui/highgui.hpp" #include "opencv2/nonfree/nonfree.hpp" #include "opencv2/legacy/legacy.hpp" using namespace cv; using namespace std; //计算原始图像点位在经过矩阵变换后在目标图像上对应位置 Point2f getTransformPoint(const Point2f originalPoint, const Mat &transformMaxtri); int main(int argc, char *argv[]) { string filename1 = "100-0038_img.jpg"; string filename2 = "100-0039_img.jpg"; Mat image01 = imread(filename1); Mat image02 = imread(filename2); imshow("拼接图像1", image01); imshow("拼接图像2", image02); //灰度图转换 Mat image1, image2; cvtColor(image01, image1, CV_RGB2GRAY); cvtColor(image02, image2, CV_RGB2GRAY); //提取特征点 SurfFeatureDetector detector(800); // surf. hessianThreshold //SiftFeatureDetector detector(800); // sift. nfeatures vector<KeyPoint> keyPoint1, keyPoint2; detector.detect(image1, keyPoint1); detector.detect(image2, keyPoint2); // 显示特征点 Mat image1KeyPoint, image2KeyPoint; drawKeypoints(image01, keyPoint1, image1KeyPoint, Scalar::all(-1), DrawMatchesFlags::DRAW_RICH_KEYPOINTS); drawKeypoints(image02, keyPoint2, image2KeyPoint, Scalar::all(-1), DrawMatchesFlags::DRAW_RICH_KEYPOINTS); imshow("image1Keypoint", image1KeyPoint); imshow("image2Keypoint", image2KeyPoint); //waitKey(0); //特征点描述，为下边的特征点匹配做准备 SurfDescriptorExtractor descriptor; //surf //SiftDescriptorExtractor descriptor; //sift Mat imageDesc1, imageDesc2; descriptor.compute(image1, keyPoint1, imageDesc1); descriptor.compute(image2, keyPoint2, imageDesc2); //获得匹配特征点，并提取最优配对 FlannBasedMatcher matcher; vector<DMatch> matchePoints; matcher.match(imageDesc1, imageDesc2, matchePoints, Mat()); // 粗匹配图像 Mat outImg; drawMatches(image01, keyPoint1, image02, keyPoint2, matchePoints, outImg, Scalar::all(-1), Scalar::all(-1), vector<char>(), DrawMatchesFlags::DRAW_RICH_KEYPOINTS); imshow("粗匹配结果", outImg); //waitKey(0); sort(matchePoints.begin(), matchePoints.end()); //特征点排序 //获取排在前N个的最优匹配特征点 vector<Point2f> imagePoints1, imagePoints2; for (int i = 0; i < 10; i++) { imagePoints1.push_back(keyPoint1[matchePoints[i].queryIdx].pt); imagePoints2.push_back(keyPoint2[matchePoints[i].trainIdx].pt); } //获取图像1到图像2的投影映射矩阵，尺寸为3*3 vector<uchar> inliersMask(imagePoints1.size()); Mat homo = findHomography(imagePoints1, imagePoints2, CV_RANSAC, 3.0, inliersMask); vector<DMatch>matches_ransac; // 手动的保留RANSAC过滤后的匹配点对  // 这一步做的没必要，因为最后确定匹配区域的时候只用到了第一个点。 for (int i = 0; i < inliersMask.size(); i++) { cout << int(inliersMask[i]) << endl; if (inliersMask[i]) { matches_ransac.push_back(matchePoints[i]); } } drawMatches(image01, keyPoint1, image02, keyPoint2, matches_ransac, outImg, Scalar::all(-1), Scalar::all(-1), vector<char>(), DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS); imshow("精匹配:", outImg); cout << "homo:\n" << homo << endl; Mat adjustMat = (Mat_<double>(3, 3) << 1.0, 0, image01.cols, 0, 1.0, 0, 0, 0, 1.0); cout << "adjustMat:\n" << adjustMat << endl; Mat adjustHomo = adjustMat * homo; cout << "adjustHomo:\n" << adjustHomo << endl; //获取最强配对点(第一个点)在原始图像和矩阵变换后图像上的对应位置，用于图像拼接点的定位 Point2f originalLinkPoint, targetLinkPoint, basedImagePoint; originalLinkPoint = keyPoint1[matchePoints[0].queryIdx].pt; //image01在image02上最强匹配点（第一个点）的对应位置 //perspectiveTransform(); targetLinkPoint = getTransformPoint(originalLinkPoint, adjustHomo); basedImagePoint = keyPoint2[matchePoints[0].trainIdx].pt; //图像配准 Mat imageTransform1; //透视变化 warpPerspective(image01, imageTransform1, adjustMat*homo, Size(image02.cols + image01.cols, image02.rows)); //imshow("imageTransform1", imageTransform1); // 因为image01已经经过变换（�