Homography

Mindmap

Video demos

42sec demo that shows computed feature correspondences & mapping of a rectangle in the right image to the image in the left using the computed homography based on the feature correspondences. You can see, that – quite often – the mapped rectangle from the right is a rectangle in the left image as well ;-)

Computing a homography using OpenCV

Here is my test using OpenCV's \opencv\bin\Release\cpp-tutorial-SURF_Homography.exe img1 img2 example to show where the left image of my office is within the right second image of my office.

SURF features are computed and matched, matches are filtered using RANSAC, and then the homography matrix H is estimated to map the image rectangle from the left into the right image (green rectangle).

Works astonishing well, despite being computing vision! :-)

The code is in \opencv\samples\cpp\tutorial_code\features2D\SURF_Homography.cpp

/**
 * @file SURF_Homography
 * @brief SURF detector + descriptor + FLANN Matcher + FindHomography
 * @author A. Huaman
 */
 
#include <stdio.h>
#include <iostream>
#include "opencv2/core/core.hpp"
#include "opencv2/features2d/features2d.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/calib3d/calib3d.hpp"
#include "opencv2/nonfree/features2d.hpp"
 
using namespace std;
using namespace cv;
 
void readme();
 
/**
 * @function main
 * @brief Main function
 */
int main( int argc, char** argv )
{
  if( argc != 3 )
  { readme(); return -1; }
 
  Mat img_object = imread( argv[1], IMREAD_GRAYSCALE );
  Mat img_scene = imread( argv[2], IMREAD_GRAYSCALE );
 
  if( !img_object.data || !img_scene.data )
  { std::cout<< " --(!) Error reading images " << std::endl; return -1; }
 
  //-- Step 1: Detect the keypoints using SURF Detector
  int minHessian = 400;
 
  SurfFeatureDetector detector( minHessian );
 
  std::vector<KeyPoint> keypoints_object, keypoints_scene;
 
  detector.detect( img_object, keypoints_object );
  detector.detect( img_scene, keypoints_scene );
 
  //-- Step 2: Calculate descriptors (feature vectors)
  SurfDescriptorExtractor extractor;
 
  Mat descriptors_object, descriptors_scene;
 
  extractor.compute( img_object, keypoints_object, descriptors_object );
  extractor.compute( img_scene, keypoints_scene, descriptors_scene );
 
  //-- Step 3: Matching descriptor vectors using FLANN matcher
  FlannBasedMatcher matcher;
  std::vector< DMatch > matches;
  matcher.match( descriptors_object, descriptors_scene, matches );
 
  double max_dist = 0; double min_dist = 100;
 
  //-- Quick calculation of max and min distances between keypoints
  for( int i = 0; i < descriptors_object.rows; i++ )
  { double dist = matches[i].distance;
    if( dist < min_dist ) min_dist = dist;
    if( dist > max_dist ) max_dist = dist;
  }
 
  printf("-- Max dist : %f \n", max_dist );
  printf("-- Min dist : %f \n", min_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_object.rows; i++ )
  { if( matches[i].distance < 3*min_dist )
    { good_matches.push_back( matches[i]); }
  }
 
  Mat img_matches;
  drawMatches( img_object, keypoints_object, img_scene, keypoints_scene,
               good_matches, img_matches, Scalar::all(-1), Scalar::all(-1),
               vector<char>(), DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS );
 
 
  //-- Localize the object from img_1 in img_2
  std::vector<Point2f> obj;
  std::vector<Point2f> scene;
 
  for( size_t i = 0; i < good_matches.size(); i++ )
  {
    //-- Get the keypoints from the good matches
    obj.push_back( keypoints_object[ good_matches[i].queryIdx ].pt );
    scene.push_back( keypoints_scene[ good_matches[i].trainIdx ].pt );
  }
 
  Mat H = findHomography( obj, scene, RANSAC );
 
  //-- Get the corners from the image_1 ( the object to be "detected" )
  std::vector<Point2f> obj_corners(4);
  obj_corners[0] = Point(0,0); obj_corners[1] = Point( img_object.cols, 0 );
  obj_corners[2] = Point( img_object.cols, img_object.rows ); obj_corners[3] = Point( 0, img_object.rows );
  std::vector<Point2f> scene_corners(4);
 
  perspectiveTransform( obj_corners, scene_corners, H);
 
 
  //-- Draw lines between the corners (the mapped object in the scene - image_2 )
  Point2f offset( (float)img_object.cols, 0);
  line( img_matches, scene_corners[0] + offset, scene_corners[1] + offset, Scalar(0, 255, 0), 4 );
  line( img_matches, scene_corners[1] + offset, scene_corners[2] + offset, Scalar( 0, 255, 0), 4 );
  line( img_matches, scene_corners[2] + offset, scene_corners[3] + offset, Scalar( 0, 255, 0), 4 );
  line( img_matches, scene_corners[3] + offset, scene_corners[0] + offset, Scalar( 0, 255, 0), 4 );
 
  //-- Show detected matches
  imshow( "Good Matches & Object detection", img_matches );
 
  waitKey(0);
 
  return 0;
}
 
/**
 * @function readme
 */
void readme()
{ std::cout << " Usage: ./SURF_Homography <img1> <img2>" << std::endl; }
 
public/homography.txt · Last modified: 2013/12/19 11:51 (external edit) · []
Recent changes RSS feed Powered by PHP Valid XHTML 1.0 Valid CSS Driven by DokuWiki