Mat tSIFT(String path)
{
Mat img = imread(path, CV_LOAD_IMAGE_GRAYSCALE);
//特征点描述符
Mat des;
if (!img.data){
std::cout << "Can't open" << std::endl;
system("Pause");
exit(0);
}
SiftFeatureDetector detector;
std::vector<KeyPoint> tSIFTkp;
detector.detect(img, tSIFTkp);
Mat img1;
drawKeypoints(img, tSIFTkp, img1, Scalar::all(-1), 4);
//FeaturesExtract
SiftDescriptorExtractor extractor;
//提取特征向量
extractor.compute(img,tSIFTkp,des);
showImg(img1);
return des;
}
Mat Pyramids::computeFeatures(const Mat &m, vector<KeyPoint> &keypoints)
{
Mat features;
SiftDescriptorExtractor ex;
ex.compute(m, keypoints, features);
return features;
}
int SIFTfeatureCalculate(Mat &img, vector<KeyPoint> &keypoints,Mat &descriptors ){
SiftFeatureDetector detector;
SiftDescriptorExtractor extractor;
detector.detect( img, keypoints );
extractor.compute( img, keypoints, descriptors );
}
void detectSiftMatchWithOpenCV(const char* img1_path, const char* img2_path, MatrixXf &match) {
Mat img1 = imread(img1_path);
Mat img2 = imread(img2_path);
SiftFeatureDetector detector;
SiftDescriptorExtractor extractor;
vector<KeyPoint> key1;
vector<KeyPoint> key2;
Mat desc1, desc2;
detector.detect(img1, key1);
detector.detect(img2, key2);
extractor.compute(img1, key1, desc1);
extractor.compute(img2, key2, desc2);
FlannBasedMatcher matcher;
vector<DMatch> matches;
matcher.match(desc1, desc2, matches);
match.resize(matches.size(), 6);
cout << "match count: " << matches.size() << endl;
for (int i = 0; i < matches.size(); i++) {
match(i, 0) = key1[matches[i].queryIdx].pt.x;
match(i, 1) = key1[matches[i].queryIdx].pt.y;
match(i, 2) = 1;
match(i, 3) = key2[matches[i].trainIdx].pt.x;
match(i, 4) = key2[matches[i].trainIdx].pt.y;
match(i, 5) = 1;
}
}
Mat Panorama::getDescriptors(vector<KeyPoint> kp){
cout << "Computing descriptors..." << endl;
SiftDescriptorExtractor extractor;
Mat descriptors;
extractor.compute(srcGray, kp, descriptors);
return descriptors;
}
// Faeture Detection and Decription
void det_desc_features(vector <Image>& images, bool flag)
{
// Detect the keypoints using SIFT Detector
SiftFeatureDetector detector(nfeatures, nOctaveLayers, contrastThreshold, edgeThreshold, sigma);
// Calculate descriptors (feature vectors)
SiftDescriptorExtractor extractor;
//// Detect the keypoints using SIFT Detector
//SurfFeatureDetector detector(500);
//// Calculate descriptors (feature vectors)
//SurfDescriptorExtractor extractor;
for (size_t i = 0; i < images.size(); i++)
{
/* Mat mask = Mat::zeros(images[i].getImg_gray().size(), images[i].getImg_gray().type());
Mat roi1(mask, cv::Rect(images[i].getImg_gray().cols - 60, 0, images[i].getImg_gray().cols - (images[i].getImg_gray().cols - 60), images[i].getImg_gray().rows));
roi1 = Scalar(255);
Mat roi2(mask, cv::Rect(0, 0, 60, images[i].getImg_gray().rows));
roi2 = Scalar(255);*/
// Feature Detection
vector <KeyPoint> tmp_keypoints;
detector.detect(images[i].getImg_gray(), tmp_keypoints);
cout << "Features detected in image #" << i << " : " << tmp_keypoints.size() << endl;
// Feature Description
Mat tmp_descriptors;
extractor.compute(images[i].getImg_gray(), tmp_keypoints, tmp_descriptors);
// Store keypoints and descriptors
images[i].setImageFeatures(tmp_keypoints, tmp_descriptors);
// Draw keypoints
Mat tmp_img_keypoints;
drawKeypoints(images[i].getImg_gray(), tmp_keypoints, tmp_img_keypoints, Scalar::all(-1), DrawMatchesFlags::DRAW_RICH_KEYPOINTS);
images[i].setImg_Keypoint(tmp_img_keypoints);
// Store img_keypoints
string str;
if (flag == 0)
{
str = "images/SIFT_Keypoints/Original_Image/";
}
else
{
str = "images/SIFT_Keypoints/Histogram_Equalazition/";
}
str.append("Image_");
str.append(to_string(images[i].getID()));
str.append("_Keypoints_detected_");
str.append(to_string(tmp_keypoints.size()));
str.append(".jpg");
imwrite(str, tmp_img_keypoints);
}
}
int sift_feature()
{
Mat img_1=imread("./samples/box.png",CV_LOAD_IMAGE_GRAYSCALE);//宏定义时CV_LOAD_IMAGE_GRAYSCALE=0，也就是读取灰度图像
Mat img_2=imread("./samples/box_in_scene.png",CV_LOAD_IMAGE_GRAYSCALE);//一定要记得这里路径的斜线方向，这与Matlab里面是相反的
if(!img_1.data || !img_2.data)//如果数据为空
{
cout<<"opencv error"<<endl;
return -1;
}
cout<<"open right"<<endl;
//第一步，用SIFT算子检测关键点
SiftFeatureDetector detector;//构造函数采用内部默认的
vector<KeyPoint> keypoints_1,keypoints_2;//构造2个专门由点组成的点向量用来存储特征点
detector.detect(img_1,keypoints_1);//将img_1图像中检测到的特征点存储起来放在keypoints_1中
detector.detect(img_2,keypoints_2);//同理
//在图像中画出特征点
Mat img_keypoints_1,img_keypoints_2;
drawKeypoints(img_1,keypoints_1,img_keypoints_1,Scalar::all(-1),DrawMatchesFlags::DEFAULT);//在内存中画出特征点
drawKeypoints(img_2,keypoints_2,img_keypoints_2,Scalar::all(-1),DrawMatchesFlags::DEFAULT);
imshow("sift_keypoints_1",img_keypoints_1);//显示特征点
imshow("sift_keypoints_2",img_keypoints_2);
//计算特征向量
SiftDescriptorExtractor extractor;//定义描述子对象
Mat descriptors_1,descriptors_2;//存放特征向量的矩阵
extractor.compute(img_1,keypoints_1,descriptors_1);//计算特征向量
extractor.compute(img_2,keypoints_2,descriptors_2);
//用burte force进行匹配特征向量
BruteForceMatcher<L2<float>>matcher;//定义一个burte force matcher对象
vector<DMatch>matches;
matcher.match(descriptors_1,descriptors_2,matches);
//绘制匹配线段
Mat img_matches;
drawMatches(img_1,keypoints_1,img_2,keypoints_2,matches,img_matches);//将匹配出来的结果放入内存img_matches中
//显示匹配线段
imshow("sift_Matches",img_matches);//显示的标题为Matches
waitKey(0);
return 0;
}
Mat compute_descriptors(Mat image, std::vector<KeyPoint> keypoints)
{
cout << "Extracting sift descriptors..." << endl;
SiftDescriptorExtractor extractor ;
Mat descriptor;
image.convertTo(image, CV_8U);
extractor.compute(image, keypoints , descriptor);
return descriptor;
}
int main()
{
//从文件中读入图像
Mat img_1 = imread("class.png");
Mat img_2 = imread("class2.png");
//如果读入图像失败
if (img_1.empty() || img_2.empty())
{
cout << "load image error" << endl;
return -1;
}
//显示图像
imshow("src image 1", img_1);
imshow("src image 2", img_2);
//第一步，用SIFT算子检测关键点
SiftFeatureDetector detector;//构造函数采用内部默认的
std::vector<KeyPoint> keypoints_1, keypoints_2;//构造2个专门由点组成的点向量用来存储特征点
detector.detect(img_1, keypoints_1);//将img_1图像中检测到的特征点存储起来放在keypoints_1中
detector.detect(img_2, keypoints_2);//同理
//在图像中画出特征点
Mat img_keypoints_1, img_keypoints_2;
drawKeypoints(img_1, keypoints_1, img_keypoints_1, Scalar::all(-1), DrawMatchesFlags::DEFAULT);//在内存中画出特征点
drawKeypoints(img_2, keypoints_2, img_keypoints_2, Scalar::all(-1), DrawMatchesFlags::DEFAULT);
imshow("sift_keypoints_1", img_keypoints_1);//显示特征点
imshow("sift_keypoints_2", img_keypoints_2);
//计算特征向量
SiftDescriptorExtractor extractor;//定义描述子对象
Mat descriptors_1, descriptors_2;//存放特征向量的矩阵
extractor.compute(img_1, keypoints_1, descriptors_1);//计算特征向量
extractor.compute(img_2, keypoints_2, descriptors_2);
//用burte force进行匹配特征向量
BruteForceMatcher<L2<float>>matcher;//定义一个burte force matcher对象
vector<DMatch>matches;
matcher.match(descriptors_1, descriptors_2, matches);
//绘制匹配线段
Mat img_matches;
drawMatches(img_1, keypoints_1, img_2, keypoints_2, matches, img_matches);//将匹配出来的结果放入内存img_matches中
//显示匹配线段
imshow("sift_Matches", img_matches);//显示的标题为Matches
cvWaitKey(0);
}
Mat computeSifts(const string& fileName)
{
const Mat input = cv::imread(fileName.c_str(), 0); //Load as grayscale
if(input.empty())
cout<<"ERROR: Image "<<fileName<<" was not read"<<endl;
Mat descriptors;
SiftFeatureDetector detector;
vector<cv::KeyPoint> keypoints;
detector.detect(input, keypoints);
SiftDescriptorExtractor extractor;
extractor.compute(input, keypoints, descriptors);
// cout<<descriptors<<endl;
return descriptors;
}
/*
* Function : doSift
* Description : Find sift points on the image
*
* path : path of the image
* container : container for sift keypoints and their descriptor
*/
void doSift(const string &path, struct SFeatures &container)
{
Mat img, des;
vector<KeyPoint> keypoints;
img = imread(path.c_str(), CV_LOAD_IMAGE_GRAYSCALE);
SiftFeatureDetector detector;
detector.detect(img, keypoints);
SiftDescriptorExtractor extractor;
extractor.compute(img, keypoints, des);
container.des = des;
container.keys = keypoints;
}
void ASiftDetector::detectAndCompute(const Mat& img, std::vector< KeyPoint >& keypoints, Mat& descriptors)
{
keypoints.clear();
descriptors = Mat(0, 128, CV_32F);
for(int tl = 1; tl < 6; tl++)
{
double t = pow(2, 0.5*tl);
for(int phi = 0; phi < 180; phi += 72.0/t)
{
std::vector<KeyPoint> kps;
Mat desc;
Mat timg, mask, Ai;
img.copyTo(timg);
affineSkew(t, phi, timg, mask, Ai);
#if 0
Mat img_disp;
bitwise_and(mask, timg, img_disp);
namedWindow( "Skew", WINDOW_AUTOSIZE );// Create a window for display.
imshow( "Skew", img_disp );
waitKey(0);
#endif
SiftFeatureDetector detector;
detector.detect(timg, kps, mask);
SiftDescriptorExtractor extractor;
extractor.compute(timg, kps, desc);
for(unsigned int i = 0; i < kps.size(); i++)
{
Point3f kpt(kps[i].pt.x, kps[i].pt.y, 1);
Mat kpt_t = Ai*Mat(kpt);
kps[i].pt.x = kpt_t.at<float>(0,0);
kps[i].pt.y = kpt_t.at<float>(1,0);
}
keypoints.insert(keypoints.end(), kps.begin(), kps.end());
descriptors.push_back(desc);
}
}
}
pair<vector<Point2f>, vector<Point2f> > computeMatching(Mat &img1, Mat &img2, vector<KeyPoint> &keypoints1, vector<KeyPoint> &keypoints2)
{
SiftDescriptorExtractor extractor;
Mat descriptors1, descriptors2;
extractor.compute(img1, keypoints1, descriptors1);
extractor.compute(img2, keypoints2, descriptors2);
BruteForceMatcher<L2<float> > matcher;
vector<DMatch> matches1_2, matches2_1;
matcher.match(descriptors1, descriptors2, matches1_2);
matcher.match(descriptors2, descriptors1, matches2_1);
pair<vector<Point2f>, vector<Point2f> > matches;
vector<DMatch> dmatchFiltrado;
double maxDistance = 90;
for (uint i=0; i < matches1_2.size(); i++) {
if (matches1_2[i].distance > maxDistance) {
continue;
}
pair<Point2f, Point2f> match1_2 = pair<Point2f, Point2f>(keypoints1[matches1_2[i].queryIdx].pt, keypoints2[matches1_2[i].trainIdx].pt);
for (uint j=0; j < matches2_1.size(); j++) {
if (matches2_1[j].distance > maxDistance) {
continue;
}
pair<Point2f, Point2f> match2_1 = pair<Point2f, Point2f>(keypoints1[matches2_1[j].trainIdx].pt, keypoints2[matches2_1[j].queryIdx].pt);
if (match1_2.first == match2_1.first && match1_2.second == match2_1.second) {
if (dmatchFiltrado.empty() || (matches.first.back() != match1_2.first || matches.second.back() != match1_2.second)) {
dmatchFiltrado.push_back(matches1_2[i]);
matches.first.push_back(match1_2.first);
matches.second.push_back(match1_2.second);
}
}
}
}
Mat img3;
drawMatches(img1, keypoints1, img2, keypoints2, dmatchFiltrado, img3);
imshow("Correspondencias", img3);
waitKey();
destroyWindow("Correspondencias");
return matches;
}
// パノラマ合成
Mat panorama(Mat src1, Mat src2, int width, int height)
{
// SIFT特徴点の検出と特徴量の計算
Mat gray1, gray2, des1, des2;
SiftFeatureDetector detector(2000);
SiftDescriptorExtractor extrator;
vector<KeyPoint> kps1, kps2;
cvtColor(src1, gray1, CV_BGR2GRAY);
cvtColor(src2, gray2, CV_BGR2GRAY);
detector.detect(gray1, kps1);
detector.detect(gray2, kps2);
extrator.compute(gray1, kps1, des1);
extrator.compute(gray2, kps2, des2);
// 特徴点の対応付け
vector<DMatch> matches;
BruteForceMatcher< L2<float> > matcher;
matcher.match(des1, des2, matches);
vector<Vec2f> pts1(matches.size());
vector<Vec2f> pts2(matches.size());
// ホモグラフィの計算
for (size_t i = 0; i < matches.size(); ++i){
pts1[i][0] = kps1[matches[i].queryIdx].pt.x;
pts1[i][1] = kps1[matches[i].queryIdx].pt.y;
pts2[i][0] = kps2[matches[i].trainIdx].pt.x;
pts2[i][1] = kps2[matches[i].trainIdx].pt.y;
}
Mat H = findHomography(pts1, pts2, CV_RANSAC);
// ホモグラフィ行列Hを用いてパノラマ合成
Mat dst;
warpPerspective(src1, dst, H, Size(width, height));
for (int y = 0; y < src1.rows; y++){
for (int x = 0; x < src1.cols; x++){
dst.at<Vec3b>(y, x) = src2.at<Vec3b>(y, x);
}
}
return dst;
}
int cv_featureDescriptor(CParamArray *pa)
{
using namespace cv;
// read image
string imageFN = svar.GetString("image", "./test.png");
Mat img = imread(imageFN);
// extract keypoints & descriptors
Ptr<FeatureDetector> detector;
SiftDescriptorExtractor extractor;
vector<KeyPoint> keypoints;
Mat descriptors;
detector = new SiftFeatureDetector;
detector->detect(img, keypoints);
extractor.compute(img, keypoints, descriptors);
// print keypoints
for(int i=0; i<keypoints.size(); i++) {
KeyPoint &p = keypoints[i];
printf("kp[%6d] x, y = %12f, %12f\n", i, p.pt.x, p.pt.y);
printf(" size = %12f, angle = %12f\n", p.size, p.angle);
printf(" response = %12f, octave = %3d, class_id = %4d\n", p.response, p.octave, p.class_id);
}
printf("\n");
// print descriptors
// type: CV_MAT_TYPE, CV_32F
printf("descriptor: \n");
printf(" cols = %d\n", descriptors.cols);
printf(" rows = %d\n", descriptors.rows);
printf(" channels = %d\n", descriptors.channels());
printf(" type = %d\n", descriptors.type());
return 0;
}
static void make_vocabulary()
{
if(flag==1)
{
return ;
}
cout<<" MAKING VOCABULARY...."<<endl;
for(int i=1; i<=20; i++)
{
cout<<" Reading File "<<i<<endl;
stringstream ss;
ss << path_People << "person_"<<setfill('0') << setw(3) << i <<".image.png";
cout<<ss.str()<<endl;
img=imread(ss.str(),0);
Mat tempp=imread(ss.str(),1);
//vector< vector<Point > > superpixel=make_superpixels(tempp);
//cout<<superpixel.size()<<" Superpixel size "<<endl;
for(int k=0; k<1; k++)
{
/* int x1=superpixel[k][0].x;
int y1=superpixel[k][0].y;
int x2=superpixel[k][1].x;
int y2=superpixel[k][1].y;
Mat newimg=Mat(x2-x1+1,y2-y1+1,0,Scalar(255,255,255));
for(int l=2; l<superpixel[k].size(); l++)
{
int x=superpixel[k][l].x;
int y=superpixel[k][l].y;
newimg.at<uchar>(x-x1,y-y1)=img.at<uchar>(x,y);
}*/
keypoints.clear();
detector.detect(img,keypoints);
detector.compute(img,keypoints,descriptor);
features_unclustered.push_back(descriptor);
}
}
cout<<"VOCABULARY BUILT...."<<endl;
cout<<endl;
}
/**
* @function main
*/
int main( int argc, char** argv )
{
ros::init(argc, argv, "object_detector");
ros::NodeHandle nh;
///subscribe to camera image topic
image_transport::ImageTransport it(nh);
image_transport::Subscriber sub = it.subscribe((string)IMAGE_TOPIC, 1, imageCallback);
///read calibration data
ifstream file (CALIBRATION_FILE);
if (!file.is_open()){
printf("ERROR: Unable to open calibration file\n");
return 2;
}
H=readCalibration(file);
//feature calculation of objct image
img_object = imread( (string)DATA_FOLDER+(string)IMAGE_NAME, CV_LOAD_IMAGE_GRAYSCALE );
//-- Step 1: Detect the keypoints using SURF Detector
SiftFeatureDetector detector;
detector.detect( img_object, keypoints_object );;
//-- Step 2: Calculate descriptors (feature vectors)
SiftDescriptorExtractor extractor;
extractor.compute( img_object, keypoints_object, descriptors_object );
//run service
ros::ServiceServer service = nh.advertiseService("vision/get_plate_position", get_plate_position);
ros::ServiceServer service1 = nh.advertiseService("vision/displayFrame",displayFrame);
ROS_INFO("ready to detect the plate");
ros::spin();
return 0;
}
int trainData() {
std:: string videoName="";
int n_frames[1000];
//create dictionary
int dict_size=100;//***
Mat features;
for(int i=1; i<no_videos; i++) {
stringstream temp;
temp<<i;
std::string no=temp.str();
videoName="C:/Rasika/trainvideos/video_"+no+".avi"; //*** path can be changed
//initialize capture
VideoCapture cap;
cap.open(videoName);
if(!cap.isOpened()) // check if we succeeded
return -1;
double count = cap.get(CV_CAP_PROP_FRAME_COUNT); //get the frame count
//create window to show image
//namedWindow("Video",1);
//cout<<count<<endl;
int jump=count/N;
int j=1;
int u=0;
if(count<10) {
jump=1;
}
int cnt=jump;
while(u<10) {
//Create matrix to store video frame
Mat image;
cap.set(CV_CAP_PROP_POS_FRAMES,cnt); //Set index to jump for particular count
bool success = cap.read(image);
if (!success) {
cout << "Cannot read frame " << endl;
break;
}
///////////Convert to gray scale/////////////
Mat gray_image;
cvtColor( image, gray_image, CV_BGR2GRAY );
////////EXTRACT INTEREST POINTS USING SIFT////
// vector of keypoints
std::vector<cv::KeyPoint> keypoints;
// Construct the SIFT feature detector object
SiftFeatureDetector sif(0.03,10.); // threshold //***
//Detect interest points
sif.detect(gray_image,keypoints);
////////IMSHOW THE FRAMES EXTRACTED///////////
//copy video stream to image
//cap>>image;
//print image to screen
//imshow("Video",image);
///////////Save the frames//////////////
stringstream temp2;
temp2<<j;
std::string no2=temp2.str();
std::string frame_name="frame"+no2+".jpg";
imwrite(frame_name,image);
//////////////Draw the keypoints////////////
/*
Mat featureImage;
// Draw the keypoints with scale and orientation information
drawKeypoints(image, // original image
keypoints, // vector of keypoints
featureImage, // the resulting image
Scalar(255,0,255), // color of the points
DrawMatchesFlags::DRAW_RICH_KEYPOINTS); //flag
//std::string name="image"+i;
imshow(frame_name, featureImage );
*/
////////////////////detect decriptors//////////////////
SiftDescriptorExtractor siftExtractor;
Mat siftDesc;
siftExtractor.compute(gray_image,keypoints,siftDesc);
features.push_back(siftDesc);//add the descriptors from each frame..to create one for a video
////////////////
//delay 33ms //***
//waitKey(33);
cnt+=jump;
j++;
u++;
///next frame for the same video
}
//store number of frames per video
n_frames[i-1]=j-1;
}
TermCriteria term(CV_TERMCRIT_ITER,100,0.001);//***
//retries number ***
int retries=1;
int flags=KMEANS_PP_CENTERS;
BOWKMeansTrainer bowTrainer(dict_size,term,retries,flags);
//cluster the feature vectors
Mat dictionary=bowTrainer.cluster(features);
//for further process
full_dictionary.push_back(dictionary);
///////////////////////////////////////////////////
FileStorage fs("full_dictionary.yml", FileStorage::WRITE);
fs << "vocabulary" << full_dictionary;
fs.release();
//Created Vocabulary
//Calculate histograms for the train videos
//idf_vector(full_dictionary);
return 0;
}
int main(int argc, char** argv)
{
if( argc < 2 )
{
printPrompt( argv[0] );
return -1;
}
initModule_nonfree();
// Get Input Data
ifstream file(argv[1]);
if ( !file.is_open() )
return false;
string str;
// Image Name
getline( file, str ); getline( file, str );
string image_name = str;
// Cloud Name
getline( file, str ); getline( file, str );
string cloud_name = str;
// width of images to be created.
getline( file, str ); getline( file, str );
int w = atoi(str.c_str());
// height of images to be created
getline( file, str ); getline( file, str );
int h = atoi(str.c_str());
// resolution of voxel grids
getline( file, str ); getline( file, str );
float r = atof(str.c_str());
// f (distance from pinhole)
getline( file, str ); getline( file, str );
float f = atof(str.c_str());
// thetax (initial rotation about X Axis of map)
getline( file, str ); getline( file, str );
float thetaX = atof(str.c_str());
// thetay (initial rotation about Y Axis of map)
getline( file, str ); getline( file, str );
float thetaY = atof(str.c_str());
// number of points to go to
getline( file, str ); getline( file, str );
float nop = atoi(str.c_str());
// Number of divisions
getline( file, str ); getline( file, str );
float divs = atoi(str.c_str());
// Number of images to return
getline( file, str ); getline( file, str );
int numtoreturn = atoi(str.c_str());
// Should we load or create photos?
getline( file, str ); getline( file, str );
string lorc =str.c_str();
// Directory to look for photos
getline( file, str ); getline( file, str );
string dir =str.c_str();
// Directory to look for kp and descriptors
getline( file, str ); getline( file, str );
string kdir =str.c_str();
// save photos?
getline( file, str ); getline( file, str );
string savePhotos =str.c_str();
file.close();
// Done Getting Input Data
map<vector<float>, Mat> imagemap;
map<vector<float>, Mat> surfmap;
map<vector<float>, Mat> siftmap;
map<vector<float>, Mat> orbmap;
map<vector<float>, Mat> fastmap;
imagemap.clear();
vector<KeyPoint> SurfKeypoints;
vector<KeyPoint> SiftKeypoints;
vector<KeyPoint> OrbKeypoints;
vector<KeyPoint> FastKeypoints;
Mat SurfDescriptors;
Mat SiftDescriptors;
Mat OrbDescriptors;
Mat FastDescriptors;
int minHessian = 300;
SurfFeatureDetector SurfDetector (minHessian);
SiftFeatureDetector SiftDetector (minHessian);
OrbFeatureDetector OrbDetector (minHessian);
FastFeatureDetector FastDetector (minHessian);
SurfDescriptorExtractor SurfExtractor;
SiftDescriptorExtractor SiftExtractor;
OrbDescriptorExtractor OrbExtractor;
if ( !fs::exists( dir ) || lorc == "c" )
{ // Load Point Cloud and render images
PointCloud<PT>::Ptr cloud (new pcl::PointCloud<PT>);
io::loadPCDFile<PT>(cloud_name, *cloud);
Eigen::Affine3f tf = Eigen::Affine3f::Identity();
tf.rotate (Eigen::AngleAxisf (thetaX, Eigen::Vector3f::UnitX()));
pcl::transformPointCloud (*cloud, *cloud, tf);
tf = Eigen::Affine3f::Identity();
tf.rotate (Eigen::AngleAxisf (thetaY, Eigen::Vector3f::UnitY()));
pcl::transformPointCloud (*cloud, *cloud, tf);
// Create images from point cloud
imagemap = render::createImages(cloud, nop, w, h, r, f);
if (savePhotos == "y")
{
for (map<vector<float>, Mat>::iterator i = imagemap.begin(); i != imagemap.end(); ++i)
{
// Create image name and storagename
string imfn = dir + "/";
string kpfn = kdir + "/";
for (int j = 0; j < i->first.size(); j++)
{
imfn += boost::to_string(i->first[j]) + " ";
kpfn += boost::to_string(i->first[j]) + " ";
}
imfn += ".jpg";
imwrite(imfn, i->second);
// Detect keypoints, add to keypoint map. Same with descriptors
SurfDetector.detect(i->second, SurfKeypoints);
SiftDetector.detect(i->second, SiftKeypoints);
OrbDetector.detect(i->second, OrbKeypoints);
FastDetector.detect(i->second, FastKeypoints);
SurfExtractor.compute(i->second, SurfKeypoints, SurfDescriptors);
SiftExtractor.compute(i->second, SiftKeypoints, SiftDescriptors);
OrbExtractor.compute(i->second, OrbKeypoints, OrbDescriptors);
SiftExtractor.compute(i->second, FastKeypoints, FastDescriptors);
// Store KP and Descriptors in yaml file.
kpfn += ".yml";
FileStorage store(kpfn, cv::FileStorage::WRITE);
write(store,"SurfKeypoints",SurfKeypoints);
write(store,"SiftKeypoints",SiftKeypoints);
write(store,"OrbKeypoints", OrbKeypoints);
write(store,"FastKeypoints",FastKeypoints);
write(store,"SurfDescriptors",SurfDescriptors);
write(store,"SiftDescriptors",SiftDescriptors);
write(store,"OrbDescriptors", OrbDescriptors);
write(store,"FastDescriptors",FastDescriptors);
store.release();
surfmap[i->first] = SurfDescriptors;
siftmap[i->first] = SiftDescriptors;
orbmap[i->first] = OrbDescriptors;
fastmap[i->first] = FastDescriptors;
}
}
}
else
{ // load images from the folder dir
// First look into the folder to get a list of filenames
vector<fs::path> ret;
const char * pstr = dir.c_str();
fs::path p(pstr);
get_all(pstr, ret);
for (int i = 0; i < ret.size(); i++)
{
// Load Image via filename
string fn = ret[i].string();
istringstream iss(fn);
vector<string> tokens;
copy(istream_iterator<string>(iss), istream_iterator<string>(), back_inserter<vector<string> >(tokens));
// Construct ID from filename
vector<float> ID;
for (int i = 0; i < 6; i++) // 6 because there are three location floats and three direction floats
ID.push_back(::atof(tokens[i].c_str()));
string imfn = dir + "/" + fn;
// Read image and add to imagemap.
Mat m = imread(imfn);
imagemap[ID] = m;
// Create Filename for loading Keypoints and descriptors
string kpfn = kdir + "/";
for (int j = 0; j < ID.size(); j++)
{
kpfn += boost::to_string(ID[j]) + " ";
}
kpfn = kpfn+ ".yml";
// Create filestorage item to read from and add to map.
FileStorage store(kpfn, cv::FileStorage::READ);
FileNode n1 = store["SurfKeypoints"];
read(n1,SurfKeypoints);
FileNode n2 = store["SiftKeypoints"];
read(n2,SiftKeypoints);
FileNode n3 = store["OrbKeypoints"];
read(n3,OrbKeypoints);
FileNode n4 = store["FastKeypoints"];
read(n4,FastKeypoints);
FileNode n5 = store["SurfDescriptors"];
read(n5,SurfDescriptors);
FileNode n6 = store["SiftDescriptors"];
read(n6,SiftDescriptors);
FileNode n7 = store["OrbDescriptors"];
read(n7,OrbDescriptors);
FileNode n8 = store["FastDescriptors"];
read(n8,FastDescriptors);
store.release();
surfmap[ID] = SurfDescriptors;
siftmap[ID] = SiftDescriptors;
orbmap[ID] = OrbDescriptors;
fastmap[ID] = FastDescriptors;
}
}
TickMeter tm;
tm.reset();
cout << "<\n Analyzing Images ..." << endl;
// We have a bunch of images, now we compute their grayscale and black and white.
map<vector<float>, Mat> gsmap;
map<vector<float>, Mat> bwmap;
for (map<vector<float>, Mat>::iterator i = imagemap.begin(); i != imagemap.end(); ++i)
{
vector<float> ID = i->first;
Mat Image = i-> second;
GaussianBlur( Image, Image, Size(5,5), 0, 0, BORDER_DEFAULT );
gsmap[ID] = averageImage::getPixSumFromImage(Image, divs);
bwmap[ID] = averageImage::aboveBelow(gsmap[ID]);
}
Mat image = imread(image_name);
Mat gsimage = averageImage::getPixSumFromImage(image, divs);
Mat bwimage = averageImage::aboveBelow(gsimage);
// cout << gsimage <<endl;
imwrite("GS.png", gsimage);
namedWindow("GSIMAGE (Line 319)");
imshow("GSIMAGE (Line 319)", gsimage);
waitKey(0);
vector<KeyPoint> imgSurfKeypoints;
vector<KeyPoint> imgSiftKeypoints;
vector<KeyPoint> imgOrbKeypoints;
vector<KeyPoint> imgFastKeypoints;
Mat imgSurfDescriptors;
Mat imgSiftDescriptors;
Mat imgOrbDescriptors;
Mat imgFastDescriptors;
SurfDetector.detect(image, imgSurfKeypoints);
SiftDetector.detect(image, imgSiftKeypoints);
OrbDetector.detect(image, imgOrbKeypoints);
FastDetector.detect(image, imgFastKeypoints);
SurfExtractor.compute(image, imgSurfKeypoints, imgSurfDescriptors);
SiftExtractor.compute(image, imgSiftKeypoints, imgSiftDescriptors);
OrbExtractor.compute(image, imgOrbKeypoints, imgOrbDescriptors);
SiftExtractor.compute(image, imgFastKeypoints, imgFastDescriptors);
tm.start();
cout << ">\n<\n Comparing Images ..." << endl;
// We have their features, now compare them!
map<vector<float>, float> gssim; // Gray Scale Similarity
map<vector<float>, float> bwsim; // Above Below Similarity
map<vector<float>, float> surfsim;
map<vector<float>, float> siftsim;
map<vector<float>, float> orbsim;
map<vector<float>, float> fastsim;
for (map<vector<float>, Mat>::iterator i = gsmap.begin(); i != gsmap.end(); ++i)
{
vector<float> ID = i->first;
gssim[ID] = similarities::getSimilarity(i->second, gsimage);
bwsim[ID] = similarities::getSimilarity(bwmap[ID], bwimage);
surfsim[ID] = similarities::compareDescriptors(surfmap[ID], imgSurfDescriptors);
siftsim[ID] = similarities::compareDescriptors(siftmap[ID], imgSiftDescriptors);
orbsim[ID] = 0;//similarities::compareDescriptors(orbmap[ID], imgOrbDescriptors);
fastsim[ID] = 0;//similarities::compareDescriptors(fastmap[ID], imgFastDescriptors);
}
map<vector<float>, int> top;
bool gotone = false;
typedef map<vector<float>, int>::iterator iter;
// Choose the best ones!
for (map<vector<float>, Mat>::iterator i = imagemap.begin(); i != imagemap.end(); ++i)
{
vector<float> ID = i->first;
int sim = /* gssim[ID] + 0.5*bwsim[ID] + */ 5*surfsim[ID] + 0.3*siftsim[ID] + orbsim[ID] + fastsim[ID];
// cout << surfsim[ID] << "\t";
// cout << siftsim[ID] << "\t";
// cout << orbsim[ID] << "\t";
// cout << fastsim[ID] << endl;
if (!gotone)
{
top[ID] = sim;
gotone = true;
}
iter it = top.begin();
iter end = top.end();
int max_value = it->second;
vector<float> max_ID = it->first;
for( ; it != end; ++it)
{
int current = it->second;
if(current > max_value)
{
max_value = it->second;
max_ID = it->first;
}
}
// cout << "Sim: " << sim << "\tmax_value: " << max_value << endl;
if (top.size() < numtoreturn)
top[ID] = sim;
else
{
if (sim < max_value)
{
top[ID] = sim;
top.erase(max_ID);
}
}
}
tm.stop();
double s = tm.getTimeSec();
cout << ">\n<\n Writing top " << numtoreturn << " images ..." << endl;
int count = 1;
namedWindow("Image");
namedWindow("Match");
namedWindow("ImageBW");
namedWindow("MatchBW");
namedWindow("ImageGS");
namedWindow("MatchGS");
imshow("Image", image);
imshow("ImageBW", bwimage);
imshow("ImageGS", gsimage);
vector<KeyPoint> currentPoints;
for (iter i = top.begin(); i != top.end(); ++i)
{
vector<float> ID = i->first;
cout << " Score: "<< i->second << "\tGrayScale: " << gssim[ID] << "\tBW: " << bwsim[ID] << " \tSURF: " << surfsim[ID] << "\tSIFT: " << siftsim[ID] << endl;
string fn = "Sim_" + boost::to_string(count) + "_" + boost::to_string(i->second) + ".png";
imwrite(fn, imagemap[ID]);
count++;
normalize(bwmap[ID], bwmap[ID], 0, 255, NORM_MINMAX, CV_64F);
normalize(gsmap[ID], gsmap[ID], 0, 255, NORM_MINMAX, CV_64F);
imshow("Match", imagemap[ID]);
imshow("MatchBW", bwmap[ID]);
imshow("MatchGS", gsmap[ID]);
waitKey(0);
}
cout << ">\nComparisons took " << s << " seconds for " << imagemap.size() << " images ("
<< (int) imagemap.size()/s << " images per second)." << endl;
return 0;
}
/************************************************************************************************
* extractFeaturesSIFT *
************************************************************************************************/
void CFeatureExtraction::extractFeaturesSIFT(
const CImage &img,
CFeatureList &feats,
unsigned int init_ID,
unsigned int nDesiredFeatures,
const TImageROI &ROI) const
{
bool usingROI = false;
if( ROI.xMin != 0 || ROI.xMax != 0 || ROI.yMin != 0 || ROI.yMax != 0 )
usingROI = true; // A ROI has been defined
// ROI can not be managed properly (yet) with these method, so we extract a subimage
// use a smart pointer so we just copy the pointer if the image is grayscale, or we'll create a new one if it was RGB:
CImage img_grayscale(img, FAST_REF_OR_CONVERT_TO_GRAY); // Was: auxImgPtr;
if( usingROI )
{
ASSERT_( ROI.xMin >= 0 && ROI.xMin < ROI.xMax && ROI.xMax < img.getWidth() && ROI.yMin >= 0 && ROI.yMax < img.getHeight() && ROI.yMin < ROI.yMax );
CImage auximg;
img_grayscale.extract_patch( auximg, ROI.xMin, ROI.yMin, ROI.xMax-ROI.xMin+1, ROI.yMax-ROI.yMin+1 ); // Subimage in "auxImg"
img_grayscale.swap(auximg);
}
switch( options.SIFTOptions.implementation )
{
// --------------------------------------------------------------------------------------
// Binary in C# -> OPTIONAL: Feature position already computed
// --------------------------------------------------------------------------------------
case CSBinary:
{
#ifdef MRPT_OS_WINDOWS
char filImg[2000],filOut[2000],filFeat[2000];
char paramImg[2000];
GetTempPathA(1000,filOut); os::strcat(filOut,1000,"temp_out.txt"); // OUTPUT FILE
GetTempPathA(1000,filImg); os::strcat(filImg,1000,"temp_img.bmp"); // INPUT IMAGE (BMP) FOR BINARY IN (C#)
bool onlyDesc = feats.size() > 0 ? true : false;
if( onlyDesc )
{
GetTempPathA(1000,filFeat); os::strcat(filFeat,1000,"temp_feats.txt"); // KEYPOINTS INPUT FILE
CMatrix listPoints(feats.size(),2);
for (size_t i= 0;i<feats.size();i++)
{
listPoints(i,0) = feats[i]->x;
listPoints(i,1) = feats[i]->y;
}
listPoints.saveToTextFile( filFeat, MATRIX_FORMAT_FIXED /*Float format*/ );
} // end if
// -------------------------------------------
// CALL TO "extractSIFT.exe"
// -------------------------------------------
img_grayscale.saveToFile( filImg );
// ------------------------------------
// Version with "CreateProcess":
// ------------------------------------
os::strcpy(paramImg,1000,"extractSIFT.exe -i"); os::strcat(paramImg,1000,filImg);
os::strcat(paramImg,1000," -f"); os::strcat(paramImg,1000,filOut);
os::strcat(paramImg,1000," -l"); os::strcat(paramImg,1000,filFeat);
// ------------------------------------
// Launch process
// ------------------------------------
bool ret = mrpt::system::launchProcess( paramImg );
if( !ret )
THROW_EXCEPTION( "[extractFeaturesSIFT] Could not launch external process... (extractSIFT.exe)" )
// Process Results
CFeatureList::iterator itFeat = feats.begin();
size_t nFeats;
CMatrix aux;
aux.loadFromTextFile( filOut );
std::cout << "[computeSiftFeatures] " << aux.getRowCount() << " features." << std::endl;
if( onlyDesc )
nFeats = feats.size();
else
{
nFeats = aux.getRowCount();
feats.resize( nFeats );
}
for( size_t i = 0;
itFeat != feats.end();
i++, itFeat++)
{
(*itFeat)->type = featSIFT;
(*itFeat)->x = usingROI ? aux(i,0) + ROI.xMin : aux(i,0);
(*itFeat)->y = usingROI ? aux(i,1) + ROI.yMin : aux(i,1);
(*itFeat)->orientation = aux(i,2);
(*itFeat)->scale = aux(i,3);
(*itFeat)->ID = init_ID + i;
// The descriptor:
aux.extractRow(i, (*itFeat)->descriptors.SIFT, 4);
}
remove(filImg);
remove(filOut);
#else
THROW_EXCEPTION("Unfortunately, this SIFT Implementation only runs in Windows OS, try Hess implementation");
#endif
break;
} // end case Binary in C#
case VedaldiBinary:
{
// --------------------------------------------------------------------------------------
// Binary by Vedaldi: NOT IMPLEMENTED YET. Input in PGM format
// --------------------------------------------------------------------------------------
#ifdef MRPT_OS_WINDOWS
THROW_EXCEPTION("Usage of Vedaldi Binary not implemented yet, please, try another one");
#else
THROW_EXCEPTION("Unfortunately, this SIFT Implementation only runs in Windows OS, try Hess implementation");
#endif
break;
} // end case Binary by Vedaldi
// --------------------------------------------------------------------------------------
// Binary by David Lowe
// --------------------------------------------------------------------------------------
case LoweBinary: // Binary by Lowe
{
#ifdef MRPT_OS_WINDOWS
char filImg[2000],filOut[2000];
char paramImg[2000];
feats.clear();
GetTempPathA(1000,filOut); os::strcat(filOut,1000,"temp_out.txt"); // OUTPUT FILE
GetTempPathA(1000,filImg); os::strcat(filImg,1000,"temp_img.pgm"); // INPUT IMAGE (PGM) FOR ORIGINAL BINARY BY LOWE
bool valid = img_grayscale.saveToFile( filImg );
if(!valid)
THROW_EXCEPTION( "An error occurred when saving input image into a .pgm file");
// CONVERT TO UNCOMPRESSED RAW PGM (TODO: Solve in a better way this issue)
os::strcpy( paramImg,1000, format( "cmd /C gmic.exe %s -o %s -quiet", filImg, filImg ).c_str() );
bool ret = mrpt::system::launchProcess( paramImg );
if(!ret)
THROW_EXCEPTION("[extractFeaturesSIFT] Could not launch external process... (gmic.exe)");
// ------------------------------------
// Version with "CreateProcess":
// ------------------------------------
os::strcpy(paramImg,1000,"cmd /C siftWin32.exe <"); os::strcat(paramImg,1000,filImg);
os::strcat(paramImg,1000," >"); os::strcat(paramImg,1000,filOut);
ret = mrpt::system::launchProcess( paramImg );
if(!ret)
THROW_EXCEPTION("[extractFeaturesSIFT] Could not launch external process... (siftWin32.exe)");
// ------------------------------------
// Process Results
// ------------------------------------
unsigned int dLen, nFeats;
FILE *f = os::fopen( filOut, "rt");
if(!f)
THROW_EXCEPTION( "Error in extract SIFT with Lowe binary, output file not found!" );
fscanf( f,"%u %u", &nFeats, &dLen); // Number of feats and length of the descriptor
for( size_t i = 0; i < nFeats; i++ )
{
CFeaturePtr feat = CFeature::Create();
feat->type = featSIFT; // Type
feat->ID = init_ID + i; // Identifier
// Position, orientation and scale
// IMPORTANTE NOTE: Lowe format stores first the 'y' coordinate and then the 'x' one
float fx,fy,fo,fs;
fscanf( f, "%f %f %f %f", &fy, &fx, &fo, &fs );
feat->x = usingROI ? fx + ROI.xMin : fx;
feat->y = usingROI ? fy + ROI.yMin : fy;
feat->orientation = fo;
feat->scale = fs;
// The descriptor
feat->descriptors.SIFT.resize( dLen );
unsigned int c;
for(unsigned int k = 0; k < dLen; k++)
{
fscanf( f, "%u", &c );
feat->descriptors.SIFT[k] = (unsigned char)c;
}
feats.push_back( feat );
} // end for
os::fclose( f );
remove(filImg);
remove(filOut);
#else
THROW_EXCEPTION("Unfortunately, this SIFT Implementation only runs in Windows OS, try Hess implementation");
#endif
break;
} // end case Binary by Lowe
// --------------------------------------------------------------------------------------
// Hess implementation
// --------------------------------------------------------------------------------------
case Hess: // Implementation by Robert Hess
{
#if !MRPT_HAS_SIFT_HESS
THROW_EXCEPTION("Method not available since MRPT has been compiled without Hess' SIFT library")
#elif MRPT_HAS_OPENCV // OK, we have Hess' sift:
IplImage* init_img;
IplImage*** gauss_pyr, *** dog_pyr;
CvMemStorage* storage;
CvSeq* features;
int octvs;
std::cout << "got to hess 1";//gb
/* check arguments */
ASSERT_(img_grayscale.getWidth() != 0 && img_grayscale.getHeight() != 0);
std::cout << "got to hess 2";//gb
/* build scale space pyramid; smallest dimension of top level is ~4 pixels */
const IplImage* ipl_im = img_grayscale.getAs<IplImage>();
std::cout << "got to hess 3"; //gb the program crashes in the next line
init_img = create_init_img( ipl_im, SIFT_IMG_DBL, SIFT_SIGMA );
std::cout << "got to hess 3b";//gb
octvs = log( (float)(MIN( init_img->width, init_img->height )) ) / log((float)2) - 2;
std::cout << "got to hess 4";//gb
gauss_pyr = build_gauss_pyr( init_img, octvs, SIFT_INTVLS, SIFT_SIGMA );
std::cout << "got to hess 5";//gb
dog_pyr = build_dog_pyr( gauss_pyr, octvs, SIFT_INTVLS );
std::cout << "got to hess 6";//gb
storage = cvCreateMemStorage( 0 );
std::cout << "got to hess 7"; //gb
features = scale_space_extrema( dog_pyr, octvs, SIFT_INTVLS,
options.SIFTOptions.threshold, // SIFT_CONTR_THR,
options.SIFTOptions.edgeThreshold, // SIFT_CURV_THR
storage );
calc_feature_scales( features, SIFT_SIGMA, SIFT_INTVLS );
if( SIFT_IMG_DBL )
adjust_for_img_dbl( features );
calc_feature_oris( features, gauss_pyr );
compute_descriptors( features, gauss_pyr, SIFT_DESCR_WIDTH, SIFT_DESCR_HIST_BINS );
/* sort features by decreasing scale and move from CvSeq to array */
cvSeqSort( features, (CvCmpFunc)feature_cmp, NULL );
/* get only the desired features */
if( nDesiredFeatures > 0 )
{
if( nDesiredFeatures < (unsigned int)features->total )
cvSeqPopMulti( features, NULL, features->total - nDesiredFeatures );
else
cout << "[Warning] Detected less features than the requested " << features->total << " vs " << nDesiredFeatures << endl;
} // end if
/* convert CvSeq into a FeatureList */
convertCvSeqInCFeatureList( features, feats, init_ID, ROI );
// clear Hess-features
cvClearSeq( features );
cvReleaseMemStorage( &storage );
cvReleaseImage( &init_img );
release_pyr( &gauss_pyr, octvs, SIFT_INTVLS + 3 );
release_pyr( &dog_pyr, octvs, SIFT_INTVLS + 2 );
#else
THROW_EXCEPTION("Method not available since MRPT has been compiled without OpenCV")
#endif //MRPT_HAS_OPENCV
break;
} // end case Hess
//***********************************************************************************************
// USING OPENCV
//***********************************************************************************************
case OpenCV:
{
#if MRPT_HAS_OPENCV && MRPT_HAS_OPENCV_NONFREE
#if MRPT_OPENCV_VERSION_NUM >= 0x211 && MRPT_OPENCV_VERSION_NUM < 0x300
SiftFeatureDetector SIFTDetector(
options.SIFTOptions.threshold, //SIFT::DetectorParams::GET_DEFAULT_THRESHOLD(),
options.SIFTOptions.edgeThreshold //SIFT::DetectorParams::GET_DEFAULT_EDGE_THRESHOLD() );
);
SiftDescriptorExtractor SIFTDescriptor;
vector<KeyPoint> cv_feats; // The OpenCV output feature list
const IplImage* cGrey = img_grayscale.getAs<IplImage>();
Mat theImg = cvarrToMat( cGrey );
SIFTDetector.detect( theImg, cv_feats );
Mat desc;
SIFTDescriptor.compute( theImg, cv_feats, desc );
//fromOpenCVToMRPT( theImg, cv_feats, desc, nDesiredFeatures, outList );
const size_t N = cv_feats.size();
unsigned int nMax = nDesiredFeatures != 0 && N > nDesiredFeatures ? nDesiredFeatures : N;
const int offset = (int)this->options.patchSize/2 + 1;
const size_t size_2 = options.patchSize/2;
const size_t imgH = img.getHeight();
const size_t imgW = img.getWidth();
unsigned int i = 0;
unsigned int cont = 0;
TFeatureID nextID = init_ID;
feats.clear();
while( cont != nMax && i != N )
{
const int xBorderInf = (int)floor( cv_feats[i].pt.x - size_2 );
const int xBorderSup = (int)floor( cv_feats[i].pt.x + size_2 );
const int yBorderInf = (int)floor( cv_feats[i].pt.y - size_2 );
const int yBorderSup = (int)floor( cv_feats[i].pt.y + size_2 );
if( options.patchSize==0 || ( (xBorderSup < (int)imgW) && (xBorderInf > 0) && (yBorderSup < (int)imgH) && (yBorderInf > 0) ) )
{
CFeaturePtr ft = CFeature::Create();
ft->type = featSIFT;
ft->ID = nextID++;
ft->x = cv_feats[i].pt.x;
ft->y = cv_feats[i].pt.y;
ft->response = cv_feats[i].response;
ft->orientation = cv_feats[i].angle;
ft->scale = cv_feats[i].size;
ft->patchSize = options.patchSize; // The size of the feature patch
ft->descriptors.SIFT.resize( 128 );
memcpy( &(ft->descriptors.SIFT[0]), &desc.data[128*i], 128*sizeof(ft->descriptors.SIFT[0]) ); // The descriptor
if( options.patchSize > 0 )
{
img.extract_patch(
ft->patch,
round( ft->x ) - offset,
round( ft->y ) - offset,
options.patchSize,
options.patchSize ); // Image patch surronding the feature
}
feats.push_back( ft );
++cont;
}
++i;
}
feats.resize( cont );
#endif
#if MRPT_OPENCV_VERSION_NUM >= 0x300
using namespace cv;
vector<KeyPoint> cv_feats;
cv::Ptr<cv::xfeatures2d::SIFT>sift = cv::xfeatures2d::SIFT::create(nDesiredFeatures,3, options.SIFTOptions.threshold, options.SIFTOptions.edgeThreshold,1.6 ); //gb
const IplImage* cGrey = img_grayscale.getAs<IplImage>();
Mat theImg = cvarrToMat(cGrey);
//SIFTDetector.detect(theImg, cv_feats);
sift->detect(theImg, cv_feats); //gb
Mat desc;
//SIFTDescriptor.compute(theImg, cv_feats, desc);
sift->compute(theImg, cv_feats, desc);
//fromOpenCVToMRPT( theImg, cv_feats, desc, nDesiredFeatures, outList );
const size_t N = cv_feats.size();
unsigned int nMax = nDesiredFeatures != 0 && N > nDesiredFeatures ? nDesiredFeatures : N;
const int offset = (int)this->options.patchSize / 2 + 1;
const size_t size_2 = options.patchSize / 2;
const size_t imgH = img.getHeight();
const size_t imgW = img.getWidth();
unsigned int i = 0;
unsigned int cont = 0;
TFeatureID nextID = init_ID;
feats.clear();
while (cont != nMax && i != N)
{
const int xBorderInf = (int)floor(cv_feats[i].pt.x - size_2);
const int xBorderSup = (int)floor(cv_feats[i].pt.x + size_2);
const int yBorderInf = (int)floor(cv_feats[i].pt.y - size_2);
const int yBorderSup = (int)floor(cv_feats[i].pt.y + size_2);
if (options.patchSize == 0 || ((xBorderSup < (int)imgW) && (xBorderInf > 0) && (yBorderSup < (int)imgH) && (yBorderInf > 0)))
{
CFeaturePtr ft = CFeature::Create();
ft->type = featSIFT;
ft->ID = nextID++;
ft->x = cv_feats[i].pt.x;
ft->y = cv_feats[i].pt.y;
ft->response = cv_feats[i].response;
ft->orientation = cv_feats[i].angle;
ft->scale = cv_feats[i].size;
ft->patchSize = options.patchSize; // The size of the feature patch
ft->descriptors.SIFT.resize(128);
memcpy(&(ft->descriptors.SIFT[0]), &desc.data[128 * i], 128 * sizeof(ft->descriptors.SIFT[0])); // The descriptor
if (options.patchSize > 0)
{
img.extract_patch(
ft->patch,
round(ft->x) - offset,
round(ft->y) - offset,
options.patchSize,
options.patchSize); // Image patch surronding the feature
}
feats.push_back(ft);
++cont;
}
++i;
}
feats.resize(cont);
#endif
#else
THROW_EXCEPTION("This method requires OpenCV >= 2.1.1 with nonfree module")
#endif
break;
} // end case OpenCV
return;
default:{break;} // end default
} // end switch
} // end extractFeaturesSIFT
int main()
{
//step1 load image
Mat img1=imread("alcatraz1.jpg");
Mat img2=imread("alcatraz2.jpg");
Mat gimg1=imread("alcatraz1.jpg",CV_LOAD_IMAGE_GRAYSCALE);
Mat gimg2=imread("alcatraz2.jpg",CV_LOAD_IMAGE_GRAYSCALE);
//cvtColor(img1,gimg1,CV_BGR2GRAY);
cout<<"compute keypoint"<<endl;
//step2 compute keypoint
SiftFeatureDetector detector;
vector<KeyPoint> kp1, kp2;
detector.detect(gimg1,kp1);
detector.detect(gimg2,kp2);
//step3 compute descriptor
SiftDescriptorExtractor extractor;
Mat descriptor1,descriptor2;
extractor.compute(gimg1,kp1,descriptor1);
extractor.compute(gimg2,kp2,descriptor2);
cout<<"compute match"<<endl;
//step4 gimg1 <-->gimg2 match
BFMatcher matcher(NORM_L2);
vector<DMatch> matches1,matches2,twoside_matches;
matcher.match(descriptor1,descriptor2,matches1);
matcher.match(descriptor2,descriptor1,matches2);
cout<<"match end"<<endl;
vector<DMatch>::iterator it1;
vector<DMatch>::iterator it2;
for(it1 = matches1.begin();it1 != matches1.end();it1++)
{ for(it2 =matches2.begin();it2 != matches2.end();it2++)
{
if((*it1).queryIdx == (*it2).trainIdx && (*it2).queryIdx == (*it1).trainIdx)
{
twoside_matches.push_back(DMatch((*it1).queryIdx,(*it1).trainIdx,(*it1).distance));
//break;
}
}
}
//step5 draw twoside_matches
Mat imgmathces;
drawMatches(gimg1,kp1,gimg2,kp2,twoside_matches,imgmathces);
//load matches keypoint
int n=twoside_matches.size();
Mat_<float> matches_kp1(3,n),matches_kp2(3,n);
// vector<DMatch>::iterator it3;
// for(it3 = twoside_matches.begin();it3 != twoside_matches.end();it3++)
// {
// matches_kp1.puch_back(kp);
// }
for(int i=0;i < twoside_matches.size();i++)
{
Point2f x1=kp1[twoside_matches[i].queryIdx].pt;
Point2f x2=kp2[twoside_matches[i].queryIdx].pt;
matches_kp1(0,i)=x1.x;
matches_kp1(1,i)=x1.y;
matches_kp1(2,i)=1;
matches_kp2(0,i)=x2.x;
matches_kp2(1,i)=x2.y;
matches_kp2(2,i)=1;
}
cout<<"save keypoints"<<endl;
FileStorage fs("points.yml",FileStorage::WRITE);
fs<<"x1"<<matches_kp1;
fs<<"x2"<<matches_kp2;
fs.release();
cout<<"show match"<<endl;
namedWindow("img1",WINDOW_NORMAL);
namedWindow("img2",WINDOW_NORMAL);
namedWindow("imgmatch",WINDOW_NORMAL);
imshow("img1",gimg1);
imshow("img2",gimg2);
imshow("imgmatch",imgmathces);
waitKey(0);
return 0;
}
void FeatureMatching(const Mat& img_1,
const Mat& img_2,
vector<KeyPoint>& keypts1,
vector<KeyPoint>& keypts2,
vector<KeyPoint>& keypts1_good,
vector<KeyPoint>& keypts2_good,
vector<DMatch>* matches,
int method)
{
Mat descriptors_1, descriptors_2;
if(method == 1) // SURF descriptor
{
double minHessian = 400;
SurfFeatureDetector detector( minHessian);
detector.detect( img_1,keypts1);
detector.detect( img_2, keypts2);
//-- Step 2: Calculate descriptors (feature vectors)
SurfDescriptorExtractor extractor;
extractor.compute( img_1,keypts1, descriptors_1 );
extractor.compute( img_2, keypts2, descriptors_2 );
//-- Draw only "good" matches
/*Mat img_matches;
drawMatches( img_1, keypts1, img_2, keypts2,
good_matches, img_matches, Scalar::all(-1), Scalar::all(-1),
vector<char>(), DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS );
//-- Show detected matches
imshow( "Feature Matches", img_matches );
waitKey(0);
destroyWindow("Feature Matches");*/
}
if(method == 2) // BRIEF descriptor
{
Ptr<FeatureDetector> detector = FeatureDetector::create("ORB"); //"BRISK"
detector->detect(img_1,keypts1);
detector->detect(img_2,keypts2);
Ptr<DescriptorExtractor> extractor = DescriptorExtractor::create("ORB");
extractor->create("ORB");
extractor->compute(img_1,keypts1, descriptors_1);
extractor->compute(img_2,keypts2, descriptors_2);
}
if(method == 3) // SIFT descriptor
{
SiftFeatureDetector detector;
detector.detect( img_1,keypts1);
detector.detect( img_2, keypts2);
//-- Step 2: Calculate descriptors (feature vectors)
SiftDescriptorExtractor extractor;
extractor.compute( img_1,keypts1, descriptors_1 );
extractor.compute( img_2, keypts2, descriptors_2 );
}
if(method == 4) // KAZE descriptor
{
/*KAZEOptions options;
options.img_width = img_1.cols;
options.img_height = img_1.rows;
KAZE evolution1(options);
evolution1.Create_Nonlinear_Scale_Space(img_1);
evolution1.Feature_Detection(keypts1);
evolution1.Feature_Description(keypts1,descriptors_1);
options.img_width = img_2.cols;
options.img_height = img_2.rows;
KAZE evolution2(options);
evolution2.Create_Nonlinear_Scale_Space(img_2);
evolution2.Feature_Detection(keypts2);
evolution2.Feature_Description(keypts2,descriptors_2);*/
}
//-- Step 3: Matching descriptor vectors using BF matcher
BFMatcher matcher(NORM_L2,true);
std::vector< DMatch > matches_;
if (matches == NULL) {
matches = &matches_;
}
matcher.match( descriptors_1, descriptors_2, *matches ); // Match the feature points
double max_dist = 0; double min_dist = 1000.0;
//-- Quick calculation of max and min distances between keypoints
for(unsigned int i = 0; i < matches->size(); i++ )
{
double dist = (*matches)[i].distance;
if( dist < min_dist ) min_dist = dist;
if( dist > max_dist ) max_dist = dist;
}
std::vector< DMatch > good_matches;
vector<KeyPoint> imgpts1_good,imgpts2_good;
if (min_dist <= 0) {
min_dist = 10.0;
}
double cutoff = 4.0*min_dist;//4.0*min_dist;
std::set<int> existing_trainIdx;
for(unsigned int i = 0; i < matches->size(); i++ )
{
if ((*matches)[i].trainIdx <= 0) {
(*matches)[i].trainIdx = (*matches)[i].imgIdx;
}
if( existing_trainIdx.find((*matches)[i].trainIdx) == existing_trainIdx.end() &&
(*matches)[i].trainIdx >= 0 && (*matches)[i].trainIdx < (int)(keypts2.size()) &&
(*matches)[i].distance > 0.0 && (*matches)[i].distance < cutoff )
{
good_matches.push_back( (*matches)[i]);
keypts1_good.push_back(keypts1[(*matches)[i].queryIdx]);
keypts2_good.push_back(keypts2[(*matches)[i].trainIdx]);
existing_trainIdx.insert((*matches)[i].trainIdx);
}
}
}
void call_my_code() {
// Mat img1_rgb = imread("/Users/sunyuyin/Desktop/img_stiching/img2_online.jpg");
// Mat img2_rgb = imread("/Users/sunyuyin/Desktop/img_stiching/img1_online.jpg");
Mat img1_rgb = imread("/Users/sunyuyin/Desktop/img_stiching/img1.JPG");
Mat img2_rgb = imread("/Users/sunyuyin/Desktop/img_stiching/img2.JPG");
if (img1_rgb.empty() || img2_rgb.empty()) {
exit(-1);
}
Mat img1, img2;
cvtColor(img1_rgb, img1, CV_RGB2GRAY);
cvtColor(img2_rgb, img2, CV_RGB2GRAY);
SiftFeatureDetector detector;
vector<KeyPoint> keypoints_img1, keypoints_img2;
detector.detect(img1, keypoints_img1);
detector.detect(img2, keypoints_img2);
SiftDescriptorExtractor extractor;
Mat descriptors_img1, descriptors_img2;
extractor.compute(img1, keypoints_img1, descriptors_img1);
extractor.compute(img2, keypoints_img2, descriptors_img2);
FlannBasedMatcher matcher;
vector<DMatch> matches;
matcher.match(descriptors_img1, descriptors_img2, matches);
double max_dist = 0.0, min_dist = numeric_limits<double>::max();
for (int i = 0; i < descriptors_img1.rows; ++i) {
double dist = matches[i].distance;
if (dist < min_dist) min_dist = dist;
if (dist > max_dist) max_dist = dist;
}
cout << "-- Max dist: " << max_dist << endl;
cout << "-- Min dist: " << min_dist << endl;
vector<DMatch> good_matches;
for (int i = 0; i < descriptors_img1.rows; ++i) {
if (matches[i].distance < 3 * min_dist) {
good_matches.push_back(matches[i]);
}
}
vector<Point2f> img1_matches;
vector<Point2f> img2_matches;
for (int i = 0; i < good_matches.size(); ++i) {
img1_matches.push_back(keypoints_img1[good_matches[i].queryIdx].pt);
img2_matches.push_back(keypoints_img2[good_matches[i].trainIdx].pt);
}
Mat H = findHomography(img1_matches, img2_matches, CV_RANSAC);
Mat result;
cout << H << endl;
warpPerspective(img1_rgb, result, H, Size(img1_rgb.cols+img2_rgb.cols,img1_rgb.rows));
/*
namedWindow("img1");
imshow("img1", img1);
namedWindow("result");
imshow("result", result);
*/
Mat half(result, Rect(0, 0, img2_rgb.cols, img2_rgb.rows));
img2_rgb.copyTo(half);
imshow("Result", result);
/*
Mat img_matches;
drawMatches(img1, keypoints_img1, img2, keypoints_img2, good_matches, img_matches, Scalar::all(-1), Scalar::all(-1),
vector<char>(), DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS);
namedWindow("img_matches");
imshow("img_matches", img_matches);
*/
/*
Mat img1_output, img2_output;
drawKeypoints(img1, keypoints_img1, img1_output);
namedWindow("Image 1 keypoints", WINDOW_AUTOSIZE);
imshow("Image 1 keypoints", img1_output);
drawKeypoints(img2, keypoints_img2, img2_output);
namedWindow("Image 2 keypoints");
imshow("Image 2 keypoints", img2_output);
*/
}
/*void normalize(Mat* srcMat, Mat* dstMat) {
//cout<<"START#";
int nRows = srcMat->rows;
int nCols = srcMat->cols;
vector<float> maxValues(nCols), minValues(nCols);
srcMat->row(0).copyTo(maxValues);
srcMat->row(0).copyTo(minValues);
for(int col = 0; col < nCols; col++) {
float max = maxValues[col];
float min = minValues[col];
float temp;
for(int row = 0; row < nRows; row++) {
temp = dstMat->at<float>(row,col);
if(temp > max)
max = temp;
if(temp < min)
min = temp;
}
maxValues[col] = max;
minValues[col] = min;
}
for(int row = 0; row < nRows; row++) {
for(int col = 0; col < nCols; col++) {
dstMat->at<float>(row,col) = float(srcMat->at<float>(row,col) - minValues[col])/float(maxValues[col] - minValues[col]);
}
}
for(int col = 0; col < nCols; col++) {
cout<<maxValues[col]<<" ";
}
cout<<"\n";
for(int col = 0; col < nCols; col++) {
cout<<minValues[col]<<" ";
}
cout<<"===================================\n";
//cout<<"END#";
}*/
int main()
{
// parameters to set
int num_classes = 10;
int PCA_FLAG = 0; // set this to use PCA
int dimensionToReduceTo = 45; // PCA good value 45,15
// output:
// SIFT_PCA_NaiveBayes_feature_train.txt
// SIFT_PCA_NaiveBayes_label_train.txt
// SIFT_PCA_NaiveBayes_feature_test.txt
// SIFT_PCA_NaiveBayes_label_test.txt
// sift_pca_bagOfwords_results.txt
if(PCA_FLAG==0)
dimensionToReduceTo = 128;
// Read parameters
string raw_data_location;
int num_training_samples_class;
int num_testing_samples_class;
vector<string> feature_train_image_names;
vector<string> feature_test_image_names;
ifstream fin;
fin.open("dataDescription.txt");
if(fin){
string temp;
getline(fin,raw_data_location);
getline(fin,temp);
num_training_samples_class = atoi(temp.c_str());
getline(fin,temp);
num_testing_samples_class = atoi(temp.c_str());
fin.close();
}
else
{
cout<<"Unable to open dataDesciption.txt\n";
exit(1);
}
// Make a list of all valid class names
string class_name_array[num_classes];
fin.open("feature_class_names.txt");
string temp;
if(fin){
vector<string> validClassNames;
while(getline(fin,temp)){
temp = temp.substr(0, temp.find("\t"));
validClassNames.push_back(temp);
}
fin.close();
if( num_classes > validClassNames.size() ){
cout<<"\nWe do not enough classes that required number of samples. \nPlease reduce the"
"number of training and/or test samples you want to use";
}
else {
for(int i = 0; i < num_classes; i++) {
class_name_array[i] = validClassNames[i];
}
fin.open("feature_train_image_names.txt");
if(fin){
string temp;
while(getline(fin,temp)){
feature_train_image_names.push_back(temp);
}
fin.close();
}
else
{
cout<<"Unable to open feature_train_image_names.txt \nPlease run randomDataSubSampler.cpp first\n";
exit(1);
}
fin.open("feature_test_image_names.txt");
if(fin){
string temp;
while(getline(fin,temp)){
feature_test_image_names.push_back(temp);
}
fin.close();
}
else
{
cout<<"Unable to open feature_test_image_names.txt \nPlease run randomDataSubSampler.cpp first\n";
exit(1);
}
}
}
else
{
cout<<"Unable to open feature_class_names.txt. \nPlease run randomDataSubSampler.cpp first\n";
exit(1);
}
// declare space to store SIFT features in 128X(total number of keypoints)
//vector< vector<double> > sift_feature_matrix;
Mat sift_feature_matrix;
// store the number of keypoints in each image
Mat_<int> num_keypoints_matrix(num_classes,num_training_samples_class);
// iterate over each class one by one
int cur_class = 0;
int cum_image_num = 0;
int labels_train[num_classes*num_training_samples_class];
for(cur_class = 0; cur_class < num_classes; cur_class++) {
string cur_class_raw_data_location = raw_data_location + class_name_array[cur_class] + "/";
for(int cur_image_num = 0; cur_image_num < num_training_samples_class; cum_image_num++, cur_image_num++) {
string cur_image_location = cur_class_raw_data_location + feature_train_image_names[cum_image_num];
// cout<<cur_image_location<<"\t";
Mat cur_image = imread(cur_image_location,0);
/* imshow("curIMage",cur_image);
waitKey(0);*/
SiftFeatureDetector detector;
vector<cv::KeyPoint> image_keypoints;
detector.detect(cur_image, image_keypoints);
/* int a = image_keypoints.size();
cout<<a<<"\t";*/
Mat tempImg;
/* drawKeypoints(cur_image, image_keypoints, tempImg);
imshow("img",tempImg);
waitKey(0);*/
num_keypoints_matrix[cur_class][cur_image_num] = image_keypoints.size();
//cout<<num_keypoints_matrix[cur_class][cur_image_num]<<"\n";
// Calculate descriptors: For each of the key points
// obtain the features describing the vicinity of the
// the key points. This will be a 128 dimensional vector
// at each key point
SiftDescriptorExtractor extractor;
Mat kepoint_descriptors;
extractor.compute( cur_image, image_keypoints, kepoint_descriptors );
sift_feature_matrix.push_back(kepoint_descriptors);
labels_train[cum_image_num] = cur_class;
}
}
// PCA to reduce dimensionality from 128 features to dimensionToReduceTo
Mat_<float> pcaSIFT_feature_matrix;
PCA pca(sift_feature_matrix, Mat(), CV_PCA_DATA_AS_ROW, dimensionToReduceTo);
if(PCA_FLAG==1){
int reducedDimension = dimensionToReduceTo;
Size size_sift_feature_matrix = sift_feature_matrix.size();
Mat_<float> projected(size_sift_feature_matrix.height,reducedDimension);
pca.project(sift_feature_matrix,projected);
projected.convertTo(pcaSIFT_feature_matrix,CV_32F);
}
else
{
pcaSIFT_feature_matrix = sift_feature_matrix;
}
// number of key points in each class
vector<int> training_totalKeyPoints_class(num_classes,0);
for(cur_class = 0; cur_class < num_classes; cur_class++) {
int sum = 0;
for(int imgNo = 0; imgNo < num_training_samples_class; imgNo++) {
training_totalKeyPoints_class[cur_class] = training_totalKeyPoints_class[cur_class] + num_keypoints_matrix.at<int>(cur_class,imgNo);
}
}
// ===============================================================
// Read Test Images
// ===============================================================
Mat_<int> testing_num_keypoints_matrix(num_classes,num_testing_samples_class);
Mat testing_sift_feature_matrix;
int cum_image_index = 0;
int testLabels[num_classes*num_testing_samples_class];
for(cur_class = 0; cur_class < num_classes; cur_class++) {
string cur_class_raw_data_location = raw_data_location + "/" + class_name_array[cur_class] + "/";
//read image of the testing data of the current_class one at a time
for(int cur_image_num = 0; cur_image_num < num_testing_samples_class; cum_image_index++,cur_image_num++) {
string cur_image_location = cur_class_raw_data_location + feature_test_image_names[cum_image_index];
Mat cur_image = imread(cur_image_location,0);
SiftFeatureDetector detector;
vector<cv::KeyPoint> image_keypoints;
detector.detect(cur_image, image_keypoints);
testing_num_keypoints_matrix[cur_class][cur_image_num] = image_keypoints.size();
// Calculate descriptors: For each of the key points
// obtain the features describing the vicinity of the
// the key points. This will be a 128 dimensional vector
// at each key point
SiftDescriptorExtractor extractor;
Mat kepoint_descriptors;
extractor.compute( cur_image, image_keypoints, kepoint_descriptors );
testing_sift_feature_matrix.push_back(kepoint_descriptors);
testLabels[cum_image_index]=cur_class;
}
}
// Project the test image SIFT feature to the PCA reduced
// dimension plane
Mat_<float> testing_pcaSIFT_feature_matrix;
if(PCA_FLAG==1){
Size size_testing_sift_feature_matrix = testing_sift_feature_matrix.size();
Mat_<float> testing_projected(size_testing_sift_feature_matrix.height,dimensionToReduceTo);
pca.project(testing_sift_feature_matrix,testing_projected);
testing_projected.convertTo(testing_pcaSIFT_feature_matrix,CV_32F);
}
else{
testing_pcaSIFT_feature_matrix = testing_sift_feature_matrix;
}
Size train_dimension = pcaSIFT_feature_matrix.size();
Size test_dimension = testing_pcaSIFT_feature_matrix.size();
ofstream fout;
// Write to file
fout.open("SIFT_PCA_NaiveBayes_feature_test.txt");
for(int i = 0; i < test_dimension.height;i++){
for(int j = 0; j < test_dimension.width; j++){
fout<<testing_pcaSIFT_feature_matrix.at<float>(i, j)<<" ";
}
fout<<"\n";
}
fout.clear();
fout.close();
fout.open("SIFT_PCA_NaiveBayes_label_test.txt");
fout.clear();
for(int i = 0; i < num_testing_samples_class*num_classes;i++){
fout<<testLabels[i]<<"\n";
}
fout.close();
fout.open("SIFT_PCA_NaiveBayes_feature_train.txt");
fout.clear();
for(int i = 0; i < train_dimension.height;i++){
for(int j = 0; j < train_dimension.width; j++){
fout<<pcaSIFT_feature_matrix.at<float>(i, j)<<" ";
}
fout<<"\n";
}
fout.close();
fout.open("SIFT_PCA_NaiveBayes_label_train.txt");
fout.clear();
for(int i = 0; i < num_training_samples_class*num_classes; i++) {
fout<<labels_train[i]<<"\n";
}
fout.close();
fout.open("SIFT_PCA_NaiveBayes_label_train.txt");
fout.clear();
for(int i = 0; i < num_training_samples_class*num_classes; i++) {
fout<<labels_train[i]<<"\n";
}
fout.close();
fout.open("SIFT_PCA_NaiveBayes_train_keypoints_matrix.txt");
fout.clear();
for(int i = 0; i < num_classes;i++){
for(int j = 0; j < num_training_samples_class; j++){
fout<<num_keypoints_matrix.at<int>(i, j)<<" ";
}
fout<<"\n";
}
fout.close();
fout.open("SIFT_PCA_NaiveBayes_test_keypoints_matrix.txt");
fout.clear();
for(int i = 0; i < num_classes;i++){
for(int j = 0; j < num_testing_samples_class; j++){
fout<<testing_num_keypoints_matrix.at<int>(i, j)<<" ";
}
fout<<"\n";
}
fout.close();
/*cout<<"---1\n";
// Construct KD Trees
vector< KDTree > kdTrees(num_classes);
int dimension = pcaSIFT_feature_matrix.size().width;
int min_keypoint_class_index = 0;
for(int curClass = 0; curClass < num_classes; curClass++) {
cout<<"\t--->"<<curClass;
int numKeyPointsCurClass = training_totalKeyPoints_class[curClass];
Mat curClassDesriptors = pcaSIFT_feature_matrix(Rect(0,min_keypoint_class_index,dimension,min_keypoint_class_index+numKeyPointsCurClass));
min_keypoint_class_index = min_keypoint_class_index + numKeyPointsCurClass;
cout<<"|";
KDTree kdCur(curClassDesriptors);
kdTrees.push_back(kdCur);
cout<<">\n";
}
cout<<"---2\n";
Mat distMat;//(num_testing_samples_class*num_classes,num_classes);
int min_keypoint_img_index = 0;
int cumImage_index = 0;
for(int curClass = 0; curClass < num_classes; curClass++) {
// int numKeyPointsCurClass = training_totalKeyPoints_class[curClass];
// Mat curClassDesriptors = pcaSIFT_feature_matrix(Rect(0,min_keypoint_class_index,dimension,min_keypoint_class_index+numKeyPointsCurClass));
// min_keypoint_class_index = min_keypoint_class_index + numKeyPointsCurClass;
KDTree kd = kdTrees[curClass];
for(int curImage = 0; curImage < num_testing_samples_class; cumImage_index++,curImage++) {
int numKeyPointsCurTestImg = num_keypoints_matrix[curClass][curImage];
Mat curTestImgDesriptorsMatrix = testing_pcaSIFT_feature_matrix(Rect(0, min_keypoint_img_index, dimension,min_keypoint_img_index+numKeyPointsCurTestImg));
min_keypoint_img_index = min_keypoint_img_index + numKeyPointsCurTestImg;
vector<int> NN_indices(curTestImgDesriptorsMatrix.size().height);
kd.findNearest(curTestImgDesriptorsMatrix,1,32,NN_indices,noArray(),noArray());
double dist_sum = 0;
for(int i = 0; i < NN_indices.size(); i++){
vector<float> ref(curTestImgDesriptorsMatrix.row(i));
vector<float> NN_point(kd.getPoint(NN_indices[i]),kd.getPoint(NN_indices[i])+dimension);
dist_sum = dist_sum + norm(ref,NN_point,NORM_L2);
}
distMat.at<float>(cumImage_index,curClass) = dist_sum;
}
}
cout<<"---3\n";
vector<int> predictedLabels(num_testing_samples_class*num_classes);
for(int i = 0 ; i < num_testing_samples_class*num_classes; i++) {
float min = distMat.at<float>(i,0);
for(int j = 0; j < num_classes; j++){
if(distMat.at<float>(i,j) < min){
min = distMat.at<float>(i,j);
predictedLabels[i] = j;
}
}
}
for(int i = 0 ; i < num_testing_samples_class*num_classes; i++) {
cout<<predictedLabels[i]<<" ";
}*/
/*
cout<<"---2\n";
Mat distMat;
int dimension = pcaSIFT_feature_matrix.size().width;
int min_keypoint_class_index = 0;
int min_keypoint_img_index = 0;
int cumImage_index = 0;
for(int curClass = 0; curClass < num_classes; curClass++) {
cout<<"\t---"<<curClass<<"\n";
int numKeyPointsCurClass = training_totalKeyPoints_class[curClass];
Mat curClassDesriptors = pcaSIFT_feature_matrix(Rect(0,min_keypoint_class_index,dimension,min_keypoint_class_index+numKeyPointsCurClass));
min_keypoint_class_index = min_keypoint_class_index + numKeyPointsCurClass;
for(int curImage = 0; curImage < num_testing_samples_class; cumImage_index++,curImage++) {
cout<<"\t\t---"<<curImage;
int numKeyPointsCurTestImg = num_keypoints_matrix[curClass][curImage];
Mat curTestImgDesriptorsMatrix = testing_pcaSIFT_feature_matrix(Rect(0, min_keypoint_img_index, dimension,min_keypoint_img_index+numKeyPointsCurTestImg));
min_keypoint_img_index = min_keypoint_img_index + numKeyPointsCurTestImg;
cout<<"D";
FlannBasedMatcher flann_matcher;
std::vector< DMatch > flann_matches;
flann_matcher.match( curTestImgDesriptorsMatrix, curClassDesriptors, flann_matches );
cout<<"F";
double dist_sum = 0;
for(int i = 0; i < numKeyPointsCurTestImg; i++){
vector<float> ref(curTestImgDesriptorsMatrix.row(i));
vector<float> NN_point(curClassDesriptors.row(flann_matches[i].trainIdx));
dist_sum = dist_sum + norm(ref,NN_point,NORM_L2);
}
distMat.at<float>(cumImage_index,curClass) = dist_sum;
cout<<">\n";
}
}
cout<<"---3\n";
vector<int> predictedLabels(num_testing_samples_class*num_classes);
for(int i = 0 ; i < num_testing_samples_class*num_classes; i++) {
float min = distMat.at<float>(i,0);
for(int j = 0; j < num_classes; j++){
if(distMat.at<float>(i,j) < min){
min = distMat.at<float>(i,j);
predictedLabels[i] = j;
}
}
}
*/
/*
// k means clustering
// labels: vector storing the labels assigned to each vector
// (the pcaSIFT feature of a keypoint). Therefore labels
// is of size = total number of keypoints = size_sift_feature_matrix.height
vector<int> labels;//(size_sift_feature_matrix.height);
int attempts = 5;
Mat centers;
TermCriteria criteria(CV_TERMCRIT_ITER|CV_TERMCRIT_EPS, 20000, 0.0001);
kmeans(pcaSIFT_feature_matrix, num_clusters, labels,criteria, attempts, KMEANS_RANDOM_CENTERS,centers );
// Object Feature Vector
// computing histograms of each image
// the keypoint_matrix stores the number of keypoints of each image
// each image has a different number of keypoints
// using this matrix, we will compute the histogram for each image
// Also, note that the pcaSIFT_matrix stores the pcaSift_features in
// following order:
// pcaSift_feature of keypoint 1 of image 1 of class 1
// pcaSift_feature of keypoint 2 of image 1 of class 1
// .
// .
// pcaSift_feature of keypoint 1 of image 2 of class 1
// pcaSift_feature of keypoint 2 of image 2 of class 1
// .
// .
// pcaSift_feature of keypoint 1 of image 1 of class 2
// .
// .
// .
// pcaSift_feature of last keypoint of last image of last class
Mat histogram_images = Mat(num_training_samples_class*num_classes, num_clusters, CV_32F, float(0.0));
vector<int> labels_train(num_training_samples_class*num_classes);
int cImg = 0;
int min_keypoint_index = 0;
int cumImage_index = 0;
for(int curClass = 0; curClass < num_classes; curClass++) {
for(int curImage = 0; curImage < num_training_samples_class; curImage++) {
int numKeypoints = num_keypoints_matrix[curClass][curImage];
for(unsigned int i = 0; i < numKeypoints; i++) {
int id = labels[min_keypoint_index+i];
histogram_images.at<float>(cumImage_index,id) += 1.0;
}
min_keypoint_index = min_keypoint_index + numKeypoints;
labels_train[cumImage_index] = curClass;
cumImage_index++;
}
}
ofstream fout;
fout.open("histogram_images.txt");
for(int i = 0; i < num_training_samples_class*num_classes;i++){
for(int j = 0; j < num_clusters; j++){
fout<<histogram_images.at<double>(i, j)<<" ";
}
fout<<"\n";
}
fout.clear();
fout.close();
// Normalize the histogram matrix
Mat normalized_histogram_images;
normalize(histogram_images, normalized_histogram_images);
histogram_images = normalized_histogram_images;
// ===============================================================
// Read Test Images
// ===============================================================
Mat_<int> testing_num_keypoints_matrix(num_classes,num_testing_samples_class);
Mat testing_sift_feature_matrix;
int cum_image_index = 0;
for(cur_class = 0; cur_class < num_classes; cur_class++) {
string cur_class_raw_data_location = raw_data_location + "/" + class_name_array[cur_class] + "/";
//read image of the testing data of the current_class one at a time
for(int cur_image_num = 0; cur_image_num < num_testing_samples_class; cum_image_index++,cur_image_num++) {
string cur_image_location = cur_class_raw_data_location + feature_test_image_names[cum_image_index];
Mat cur_image = imread(cur_image_location,0);
SiftFeatureDetector detector;
vector<cv::KeyPoint> image_keypoints;
detector.detect(cur_image, image_keypoints);
testing_num_keypoints_matrix[cur_class][cur_image_num] = image_keypoints.size();
// Calculate descriptors: For each of the key points
// obtain the features describing the vicinity of the
// the key points. This will be a 128 dimensional vector
// at each key point
SiftDescriptorExtractor extractor;
Mat kepoint_descriptors;
extractor.compute( cur_image, image_keypoints, kepoint_descriptors );
testing_sift_feature_matrix.push_back(kepoint_descriptors);
}
}
// Project the test image SIFT feature to the PCA reduced
// dimension plane
Size size_testing_sift_feature_matrix = testing_sift_feature_matrix.size();
Mat_<float> testing_projected(size_testing_sift_feature_matrix.height,reducedDimension);
pca.project(testing_sift_feature_matrix,testing_projected);
Mat_<float> testing_pcaSIFT_feature_matrix;
testing_projected.convertTo(testing_pcaSIFT_feature_matrix,CV_32F);
Mat testing_histogram_images = Mat(num_testing_samples_class*num_classes, num_clusters, CV_32F, float(0.0));
vector<int> labels_test(num_testing_samples_class*num_classes);
cImg = 0;
min_keypoint_index = 0;
cumImage_index = 0;
for(int curClass = 0; curClass < num_classes; curClass++) {
for(int curImage = 0; curImage < num_testing_samples_class; curImage++) {
int numKeypoints = testing_num_keypoints_matrix[curClass][curImage];
Mat tempDescriptor=testing_pcaSIFT_feature_matrix(cv::Rect(0,min_keypoint_index,reducedDimension,numKeypoints));
FlannBasedMatcher flann_matcher;
std::vector< DMatch > flann_matches;
flann_matcher.match( tempDescriptor, centers, flann_matches );
for(unsigned int i = 0; i < flann_matches.size(); i++) {
int id = flann_matches[i].trainIdx;
testing_histogram_images.at<float>(cumImage_index,id) += 1.0;
}
min_keypoint_index = min_keypoint_index + numKeypoints;
labels_test[cumImage_index] = curClass;
cumImage_index++;
}
}
// NORMALIZE HISTOGRAMS
Mat normalized_testing_histogram_images;
normalize(testing_histogram_images,normalized_testing_histogram_images);
testing_histogram_images = normalized_testing_histogram_images;
cout<<"\n\n===========BOW=======================\n\n";
FlannBasedMatcher flann_matcher;
vector< vector < DMatch > > flann_matches;
Mat_<float> testHist = testing_histogram_images;
Mat_<float> trainHist = histogram_images;
flann_matcher.knnMatch( testHist, trainHist, flann_matches, k_nearest_neighbor );
int predTestLabels[num_testing_samples_class*num_classes];
for(int imgNo = 0; imgNo < num_testing_samples_class*num_classes; imgNo++) {
vector < DMatch > temp = flann_matches[imgNo];
float votes[num_clusters]={0.0};
const int N = sizeof(votes) / sizeof(float);
for(int neigh = 0; neigh < temp.size(); neigh++ ) {
int id = temp[neigh].trainIdx;
int ind = id;
id = ind/num_training_samples_class;
if(ind%num_training_samples_class == 0)
id = id - 1;
float dist = temp[neigh].distance;
votes[id] = votes[id] + (1.0/dist);
}
predTestLabels[imgNo] = distance(votes, max_element(votes, votes + N));
}
// compute error
vector<float> error(num_classes,0.0);
float totalError=0.0;
for(int i = 0; i < num_testing_samples_class*num_classes; i++) {
if(predTestLabels[i] != labels_test[i])
{
error[labels_test[i]] = error[labels_test[i]] + 1.0;
totalError = totalError + 1.0;
}
}
// ofstream fout;
// Write to file
fout.open("feature_test.txt");
for(int i = 0; i < num_testing_samples_class*num_classes;i++){
for(int j = 0; j < num_clusters; j++){
fout<<testing_histogram_images.at<float>(i, j)<<" ";
}
fout<<"\n";
}
fout.clear();
fout.close();
fout.open("label_test.txt");
fout.clear();
for(int i = 0; i < num_testing_samples_class*num_classes;i++){
fout<<labels_test[i]<<"\n";
}
fout.close();
fout.open("feature_train.txt");
fout.clear();
for(int i = 0; i < num_training_samples_class*num_classes;i++){
for(int j = 0; j < num_clusters; j++){
fout<<histogram_images.at<float>(i, j)<<" ";
}
fout<<"\n";
}
fout.close();
fout.open("label_train.txt");
fout.clear();
for(int i = 0; i < num_training_samples_class*num_classes; i++) {
fout<<labels_train[i]<<"\n";
}
fout.close();
fout.open("predictedLabels.txt");
fout.clear();
for(int i = 0; i < num_testing_samples_class*num_classes; i++) {
fout<<predTestLabels[i]<<"\n";
}
fout.clear();
fout<<"\nClass Wise Number of Miss-classifications ("<<num_testing_samples_class<<" test samples "
"in each class)\n";
for(int i = 0; i < num_classes; i++) {
fout<<class_name_array[i]<<"\t: "<<error[i]<<"\n";
}
fout<<"Total Error(%)\n"<<totalError*100/(num_testing_samples_class*num_classes);
fout.close();
*/
}
int main(int argc, char* argv[])
{
int DICTIONARY_BUILD = 3;
if (DICTIONARY_BUILD == 1){
//Step 1 - Obtain the set of bags of features.
//to store the input file names
char * filename = new char[100];
//to store the current input image
Mat input;
//To store the keypoints that will be extracted by SIFT
vector<KeyPoint> keypoints;
//To store the SIFT descriptor of current image
Mat descriptor;
//To store all the descriptors that are extracted from all the images.
Mat featuresUnclustered;
//The SIFT feature extractor and descriptor
SiftDescriptorExtractor detector;
//I select 20 (1000/50) images from 1000 images to extract feature descriptors and build the vocabulary
int startid = 1;
int endid = 39;
for(int f=startid;f<=endid;f++){
//create the file name of an image
sprintf(filename,".\\Release\\omocha_train\\%i.jpg",f);
//open the file
input = imread(filename, CV_LOAD_IMAGE_GRAYSCALE); //Load as grayscale
//detect feature points
detector.detect(input, keypoints);
//compute the descriptors for each keypoint
detector.compute(input, keypoints,descriptor);
//put the all feature descriptors in a single Mat object
featuresUnclustered.push_back(descriptor);
//print the percentage
printf("%i percent done\n",f);
}
//Construct BOWKMeansTrainer
//the number of bags
int dictionarySize=200;
//define Term Criteria
TermCriteria tc(CV_TERMCRIT_ITER,100,0.001);
//retries number
int retries=1;
//necessary flags
int flags=KMEANS_PP_CENTERS;
//Create the BoW (or BoF) trainer
BOWKMeansTrainer bowTrainer(dictionarySize,tc,retries,flags);
//cluster the feature vectors
Mat dictionary=bowTrainer.cluster(featuresUnclustered);
//store the vocabulary
FileStorage fs(".\\dictionary.yml", FileStorage::WRITE);
fs << "vocabulary" << dictionary;
fs.release();
}else if(DICTIONARY_BUILD == 2){
//Step 2 - Obtain the BoF descriptor for given image/video frame.
//prepare BOW descriptor extractor from the dictionary
Mat dictionary;
FileStorage fs(".\\dictionary.yml", FileStorage::READ);
fs["vocabulary"] >> dictionary;
fs.release();
//create a nearest neighbor matcher
Ptr<DescriptorMatcher> matcher(new FlannBasedMatcher);
//create Sift feature point extracter
Ptr<FeatureDetector> detector(new SiftFeatureDetector());
//create Sift descriptor extractor
Ptr<DescriptorExtractor> extractor(new SiftDescriptorExtractor);
//create BoF (or BoW) descriptor extractor
BOWImgDescriptorExtractor bowDE(extractor,matcher);
//Set the dictionary with the vocabulary we created in the first step
bowDE.setVocabulary(dictionary);
//To store the image file name
char * filename = new char[100];
char * inputfile = new char[100];
//To store the image tag name - only for save the descriptor in a file
//char * imageTag = new char[10];
int startid = 1;
int endid = 39;
for(int i = startid; i <= endid; i++)
{
sprintf(inputfile,".\\Release\\omocha_train\\%i.jpg", i);
sprintf(filename, ".\\%i.yml", i);
//open the file to write the resultant descriptor
FileStorage fs1(filename, FileStorage::WRITE);
//read the image
Mat img=imread(inputfile,CV_LOAD_IMAGE_GRAYSCALE);
//To store the keypoints that will be extracted by SIFT
vector<KeyPoint> keypoints;
//Detect SIFT keypoints (or feature points)
detector->detect(img,keypoints);
//To store the BoW (or BoF) representation of the image
Mat bowDescriptor;
//extract BoW (or BoF) descriptor from given image
bowDE.compute(img,keypoints,bowDescriptor);
//prepare the yml (some what similar to xml) file
//sprintf(imageTag,"img1");
//write the new BoF descriptor to the file
//fs1 << imageTag << bowDescriptor;
fs1 << "imageData" << bowDescriptor;
//You may use this descriptor for classifying the image.
//release the file storage
fs1.release();
}
}else{
int main(int argc,char** argv){
Mat image1,image2;
const char* source_window = "Source image";
/// Load images
image1 = imread( argv[1], 1 );
image2 = imread( argv[2], 1 );
if( argc != 3 || !image1.data || !image2.data)
{
printf( "No image data \n" );
return 1;
}
int cols=image1.cols;
int rows=image1.rows;
// cout<<"\ntaille de la matrice:" <<image1.size();
// cout<<"\ntype de la matrice: \n" << image1.type();
// cout<<"\nflags" << image1.flags;
// cout<<"\ndims" << image1.dims;
cout<<"\nrows" << image1.rows;
cout<<"\ncols" << image1.cols;
// Point pt = Point(1,2);
// cout<<"\nnombre de chanels: " << image1.channels();
// cout<< "\npoints 1 1 " << (int)image1.at<cv::Vec3b>(0,1)[1];
/*
for(int i=0;i<cols;i++){
for(int j=0;j<rows;j++){
image1.at<cv::Vec3b>(i,j)[0]=0;
image1.at<cv::Vec3b>(i,j)[1]=0;
image1.at<cv::Vec3b>(i,j)[2]=0;
}
}
*/
cout<< "\nmais que se passe-t'il?";
// cout<<"\nimage1" << image1;
/// vector of keypoints
vector<KeyPoint> keypoints1,keypoints2;
///Construct the SURF feature detector object
SiftFeatureDetector sift;
sift.detect(image1,keypoints1);
sift.detect(image2,keypoints2);
namedWindow( "Image 1", CV_WINDOW_AUTOSIZE );
imshow( "Image 1", image1 );
namedWindow( "Image 2", CV_WINDOW_AUTOSIZE );
imshow( "Image 2", image2 );
//afficher les coordonées des points des keypoints
/*for(int i=0;i<keypoints1.size();i++){
cout<<"\n\nkeypoints number" << i <<"\n";
cout<<"\nkeypoints1" << keypoints1[i].pt;
cout<<"\nkeypoints1x " << keypoints1[i].pt.x;
cout<<"\nkeypoints1y " << keypoints1[i].pt.y;
}*/
/*Mat imcopy;
image1.copyTo(imcopy);
for(int i=0;i<keypoints1.size();i++){
imcopy.at<cv::Vec3b>(keypoints1[i].pt.y,keypoints1[i].pt.x)[0]=0;
imcopy.at<cv::Vec3b>(keypoints1[i].pt.y,keypoints1[i].pt.x)[1]=0;
imcopy.at<cv::Vec3b>(keypoints1[i].pt.y,keypoints1[i].pt.x)[2]=255;
}
namedWindow( "Image copy", CV_WINDOW_AUTOSIZE );
imshow( "Image copy", imcopy );
*/
cout << "\ntaille du vecteur de keypoints: " << keypoints1.size();
SiftDescriptorExtractor siftDesc;
Mat descriptors1,descriptors2;
siftDesc.compute(image1,keypoints1,descriptors1);
siftDesc.compute(image2,keypoints2,descriptors2);
// Construction of the matcher
BruteForceMatcher<L2<float> > matcher;
// Match the two image descriptors
vector<DMatch> matches;
matcher.match(descriptors1,descriptors2, matches);
nth_element(matches.begin(), // initial position
matches.begin()+24, // position of the sorted element
matches.end()); // end position
// remove all elements after the 25th
//display the element attributs
//cout<< "\nmatches " << matches;
//afficher les matches
for(int i=0;i<matches.size();i++){
//affichage des attributs
/* cout<< "\n\npoint num " << i;
cout<< "\nimgIdx " << matches[i].imgIdx ;
cout<< "\nqueryIdx " << matches[i].queryIdx;
cout<< "\ntrainIdx " << matches[i].trainIdx;
cout<< "\ndistance " << matches[i].distance;
*/
/*
while(matches[i].distance >100 && i<matches.size()){
cout << "\ni= " << i;
matches.erase(matches.begin()+i, matches.begin()+i+1);
}
*/
}
for(int i=0;i<matches.size();i++){
cout<< "\nOn relie le point de coordonee x1= " << keypoints1[matches[i].queryIdx].pt.x;
cout<< "\ny1= " << keypoints1[matches[i].queryIdx].pt.y;
cout<< "\nAvec le point de coordonne x2= " << keypoints2[matches[i].trainIdx].pt.x;
cout<< "\ny2= " << keypoints2[matches[i].trainIdx].pt.y;
}
cout << '\n' << "nombre de correspondances:" << matches.size() << '\n';
//matches.erase(matches.begin(), matches.end());
//keypoints1.erase(keypoints1.begin(), keypoints1.end());
//keypoints2.erase(keypoints2.begin(), keypoints2.end());
//matches.erase(matches.begin(), matches.begin()+1600);
Mat imageMatches;
Mat matchesMask;
drawMatches(
image1,keypoints1, // 1st image and its keypoints
image2,keypoints2, // 2nd image and its keypoints
matches, // the matches
imageMatches, // the image produced
Scalar::all(-1), // color of the lines
Scalar(255,255,255) //color of the keypoints
);
namedWindow( "Matches SIFT", CV_WINDOW_AUTOSIZE );
imshow( "Matches SIFT", imageMatches );
imwrite("resultat.png", imageMatches);
/*
drawKeypoints(src,keypoints1,dst,cv::Scalar(255,255,255));
cout << '\n' << keypoints1.size() << '\n';
imshow( "Image 1", dst );
imwrite("resultat.png", dst);
*/
waitKey(0);
return 0;
}
//--------------------------------------【main( )函数】-----------------------------------------
// 描述：控制台应用程序的入口函数，我们的程序从这里开始执行
//-----------------------------------------------------------------------------------------------
int main()
{
//【0】改变console字体颜色
system("color 5F");
ShowHelpText();
//【1】载入图像、显示并转化为灰度图
Mat trainImage = imread("1.jpg"), trainImage_gray;
imshow("原始图",trainImage);
cvtColor(trainImage, trainImage_gray, CV_BGR2GRAY);
//【2】检测SIFT关键点、提取训练图像描述符
vector<KeyPoint> train_keyPoint;
Mat trainDescription;
SiftFeatureDetector featureDetector;
featureDetector.detect(trainImage_gray, train_keyPoint);
SiftDescriptorExtractor featureExtractor;
featureExtractor.compute(trainImage_gray, train_keyPoint, trainDescription);
// 【3】进行基于描述符的暴力匹配
BFMatcher matcher;
vector<Mat> train_desc_collection(1, trainDescription);
matcher.add(train_desc_collection);
matcher.train();
//【4】创建视频对象、定义帧率
VideoCapture cap(0);
unsigned int frameCount = 0;//帧数
//【5】不断循环，直到q键被按下
while(char(waitKey(1)) != 'q')
{
//<1>参数设置
double time0 = static_cast<double>(getTickCount( ));//记录起始时间
Mat captureImage, captureImage_gray;
cap >> captureImage;//采集视频到testImage中
if(captureImage.empty())
continue;
//<2>转化图像到灰度
cvtColor(captureImage, captureImage_gray, CV_BGR2GRAY);
//<3>检测SURF关键点、提取测试图像描述符
vector<KeyPoint> test_keyPoint;
Mat testDescriptor;
featureDetector.detect(captureImage_gray, test_keyPoint);
featureExtractor.compute(captureImage_gray, test_keyPoint, testDescriptor);
//<4>匹配训练和测试描述符
vector<vector<DMatch> > matches;
matcher.knnMatch(testDescriptor, matches, 2);
// <5>根据劳氏算法（Lowe's algorithm），得到优秀的匹配点
vector<DMatch> goodMatches;
for(unsigned int i = 0; i < matches.size(); i++)
{
if(matches[i][0].distance < 0.6 * matches[i][1].distance)
goodMatches.push_back(matches[i][0]);
}
//<6>绘制匹配点并显示窗口
Mat dstImage;
drawMatches(captureImage, test_keyPoint, trainImage, train_keyPoint, goodMatches, dstImage);
imshow("匹配窗口", dstImage);
//<7>输出帧率信息
cout << "\t>当前帧率为：" << getTickFrequency() / (getTickCount() - time0) << endl;
}
return 0;
}
int main(int argc, char* argv[])
{
char *filename = new char[100];
vector<string> validFormats;
validFormats.push_back("png");
validFormats.push_back("ppm");
validFormats.push_back("jpg");
validFormats.push_back("gif");
validFormats.push_back("bmp");
validFormats.push_back("tiff");
int minHessian = 400; //Hessian Threshold
Mat input;
//To store the keypoints that will be extracted by SIFT
vector<KeyPoint> keypoints;
//To store the SIFT descriptor of current image
Mat descriptor;
//To store all the descriptors that are extracted from all the images.
Mat featuresUnclustered;
//The SIFT feature extractor and descriptor
SiftDescriptorExtractor detector;
DIR *dir;
struct dirent *ent;
if((dir = opendir(argv[1])) != NULL)
{
while((ent = readdir(dir)) != NULL)
{
if(ent->d_type == DT_REG)
{
string fullname(ent->d_name);
int lastindex = fullname.find_last_of(".");
string format = fullname.substr(lastindex + 1, fullname.length() - 1);
if(find(validFormats.begin(), validFormats.end(), format) != validFormats.end())
{
sprintf(filename, "%s/%s",argv[1], ent->d_name);
printf("%s\n", filename);
input = imread(filename, CV_LOAD_IMAGE_GRAYSCALE);
detector.detect(input, keypoints);
detector.compute(input, keypoints, descriptor);
featuresUnclustered.push_back(descriptor);
}
}
}
closedir(dir);
}
else
{
perror("");
return EXIT_FAILURE;
}
int dictionarySize = 200;
TermCriteria tc(CV_TERMCRIT_ITER, 100, 0.001);
int retries = 1;
int flags = KMEANS_RANDOM_CENTERS;
BOWKMeansTrainer bowTrainer(dictionarySize,tc,retries,flags);
//cout << "I'm here too\n";
Mat dictionary = bowTrainer.cluster(featuresUnclustered);
sprintf(filename, "%s/dictionary.yml", argv[2]);
FileStorage fs(filename, FileStorage::WRITE);
fs << "vocabulary" << dictionary;
fs.release();
//create a nearest neighbor matcher
Ptr<DescriptorMatcher> matcher(new FlannBasedMatcher);
//create Sift feature point extracter
Ptr<FeatureDetector> siftdetector(new SiftFeatureDetector());
//create Sift descriptor extractor
Ptr<DescriptorExtractor> extractor(new SiftDescriptorExtractor);
//create BoF (or BoW) descriptor extractor
BOWImgDescriptorExtractor bowDE(extractor,matcher);
//Set the dictionary with the vocabulary we created in the first step
bowDE.setVocabulary(dictionary);
//To store the image file name
char *filename2 = new char[100];
//To store the image tag name - only for save the descriptor in a file
char *imageTag = new char[100];
int i = 1;
if((dir = opendir(argv[1])) != NULL)
{
while((ent = readdir(dir)) != NULL)
{
if(ent->d_type == DT_REG)
{
sprintf(filename, "%s/%s",argv[1], ent->d_name);
string fullname(ent->d_name);
int lastindex = fullname.find_last_of(".");
string format = fullname.substr(lastindex + 1, fullname.length() - 1);
if(find(validFormats.begin(), validFormats.end(), format) != validFormats.end())
{
string rawname = fullname.substr(0, lastindex);
string complete = rawname + "*" + format;
// printf("Complete filename: %s\n", complete.c_str());
Mat img = imread(filename,CV_LOAD_IMAGE_GRAYSCALE);
vector<KeyPoint> keypoints;
siftdetector->detect(img,keypoints);
Mat bowDescriptor;
bowDE.compute(img,keypoints,bowDescriptor);
sprintf(filename2, "%s/siftdescriptors/%s.yml", argv[2], complete.c_str());
FileStorage fs1(filename2, FileStorage::WRITE);
printf("%s\n", filename2);
fs1 << rawname.c_str() << bowDescriptor;
fs1.release();
}
}
}
closedir(dir);
}
else
{
perror("");
return EXIT_FAILURE;
}
}
void Detect( Mat& img_scene ){
LOGI("starting object detection");
detector1.detect( img_scene, keypoints_scene );
LOGI("Keypoints detected");
extractor.compute( img_scene, keypoints_scene, descriptors_scene );
LOGI("Descriptors extracted");
FlannBasedMatcher matcher;
std::vector< DMatch > matches;
matcher.match( descriptors_source, descriptors_scene, matches );
LOGI("Matching done");
//-- Quick calculation of max and min distances between keypoints
double min_dist=1000, max_dist;
for( int i = 0; i < descriptors_source.rows; i++ )
{ double dist = matches[i].distance;
if( dist < min_dist ) min_dist = dist;
if( dist > max_dist ) max_dist = dist;
}
//-- Draw only "good" matches (i.e. whose distance is less than 3*min_dist )
std::vector< DMatch > good_matches;
for( int i = 0; i < descriptors_source.rows; i++ )
{ if( matches[i].distance <= 4*min_dist )
{ good_matches.push_back( matches[i]); }
}
// GEOM FILTER
good_matches.clear();
vector<uchar> inliers;
vector<Point2f> pts1, pts2;
for (int i = 0; i < matches.size(); i++) {
pts1.push_back(keypoints_source[matches[i].queryIdx].pt);
pts2.push_back(keypoints_scene[matches[i].trainIdx].pt);
}
Mat F = findFundamentalMat(Mat(pts1), Mat(pts2),
FM_RANSAC, 3, 0.99, inliers);
for (int i = 0; i < inliers.size(); i++) {
if ( (int)inliers[i] ) {
good_matches.push_back(matches[i]);
}
}
//-- Localize the object
std::vector<Point2f> obj;
std::vector<Point2f> scene;
for( int i = 0; i < good_matches.size(); i++ )
{
//-- Get the keypoints from the good matches
obj.push_back( keypoints_source[ good_matches[i].queryIdx ].pt );
scene.push_back( keypoints_scene[ good_matches[i].trainIdx ].pt );
}
LOGI("Point Correspondence done");
Mat img_matches;
Mat img_object = imread("/sdcard/charminarAR/obj.jpg");
drawMatches( img_object, keypoints_source, img_scene, keypoints_scene,
good_matches, img_matches, Scalar::all(-1), Scalar::all(-1),
vector<char>(), DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS );
imwrite("/sdcard/charminarAR/matches2.jpg", img_matches);
LOGI("saved matches");
prev_scene_points = source_points;
HomographicTransformation( obj, scene );
points[1] = scene;
}
vector<Mat> make_descriptors(vector< vector<Point> > superpixel,Mat img)
{
cout<<" Making Bowdescriptors "<<endl;
vector<Mat> ret;
vector<KeyPoint>keypoint1;
Mat bowdescriptor1;
//imshow("sf" , img);
//while(waitKey()!=27);
for(int k=0; k<superpixel.size(); k++)
{
int x1=superpixel[k][0].x;
int y1=superpixel[k][0].y;
int x2=superpixel[k][1].x;
int y2=superpixel[k][1].y;
Mat newimg=Mat(x2-x1+1,y2-y1+1,0,Scalar(255,255,255));
for(int l=2; l<superpixel[k].size(); l++)
{
int x=superpixel[k][l].x;
int y=superpixel[k][l].y;
newimg.at<uchar>(x-x1,y-y1)=img.at<uchar>(x,y);
}
//keypoint1.clear();
detector.detect(newimg,keypoint1);
bowde.compute(newimg,keypoint1,bowdescriptor1);
// cout<<k<<" "<<endl;
ret.push_back(bowdescriptor1);
}
for(int i=0; i<superpixel.size(); i++)
{
int cnt=1;
for(int j=0; j<superpixel.size(); j++)
{
if(i==j)
{
continue;
}
if(GR[i][j]<=N && ret[j].rows!=0 && ret[i].rows!=0)
{
if(ret[i].rows==0)
{
ret[i]=ret[j];
continue;
}
ret[i]=ret[i]+ret[j];
cnt++;
}
}
ret[i]=ret[i]/cnt;
}
cout<<" GRAPH "<<endl;
for(int i=0; i<40; i++)
{
for(int j=0; j<40; j++)
{
cout<<GR[i][j]<<" ";
}
cout<<endl;
}
cout<<endl;
cout<<" LEAVING bowdescriptors "<<endl;
return ret;
}
