Estoy tratando de detectar un objeto del siguiente código que involucra el detector de surf, no quiero dibujar coincidencias, quiero dibujar un rectángulo alrededor del objeto detectado, pero de alguna manera no puedo obtener la homografía correcta. Me voy mal.

#include <stdio.h>
#include <iostream>
#include "opencv2/core/core.hpp"
#include "opencv2/features2d/features2d.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/calib3d/calib3d.hpp"

using namespace cv;

int main()
{
    Mat object = imread( "sample.jpeg", CV_LOAD_IMAGE_GRAYSCALE );

    if( !object.data )
    {
        std::cout<< "Error reading object " << std::endl;
        return -1;
    }

    //Detect the keypoints using SURF Detector
    int minHessian = 500;

    SurfFeatureDetector detector( minHessian );
    std::vector<KeyPoint> kp_object;

    detector.detect( object, kp_object );

    //Calculate descriptors (feature vectors)
    SurfDescriptorExtractor extractor;
    Mat des_object;

    extractor.compute( object, kp_object, des_object );

    FlannBasedMatcher matcher;

    VideoCapture cap(0);

    namedWindow("Good Matches",0);
    cvResizeWindow("Good Matches",800,800);

    std::vector<Point2f> obj_corners(4);

    //Get the corners from the object
    obj_corners[0] = (cvPoint(0,0));
    obj_corners[1] = (cvPoint(object.cols,0));
    obj_corners[2] = (cvPoint(object.cols,object.rows));
    obj_corners[3] = (cvPoint(0, object.rows));

    char key = 'a';
    int framecount = 0;
    while (key != 27)
    {
        Mat frame;
        cap >> frame;

        if (framecount < 5)
        {
            framecount++;
            continue;
        }

        Mat des_image, img_matches;
        std::vector<KeyPoint> kp_image;
        std::vector<vector<DMatch > > matches;
        std::vector<DMatch > good_matches;
        std::vector<Point2f> obj;
        std::vector<Point2f> scene;
        std::vector<Point2f> scene_corners(4);
        Mat H;
        Mat image;

        cvtColor(frame, image, CV_RGB2GRAY);

        detector.detect( image, kp_image );
       extractor.compute( image, kp_image, des_image );

        matcher.knnMatch(des_object, des_image, matches, 2);

       for(int i = 0; i < min(des_image.rows-1,(int) matches.size()); i++) //THIS LOOP IS SENSITIVE TO SEGFAULTS
      {
            if((matches[i][0].distance < 0.6*(matches[i][1].distance)) && ((int) matches[i].size()<=2 && (int) matches[i].size()>0))
         {
              good_matches.push_back(matches[i][0]);
            }
       }

        //Draw only "good" matches
     //  drawMatches( object, kp_object, image, kp_image, good_matches, img_matches, Scalar::all(-1), Scalar::all(-1), vector<char>(),DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS);

        if (good_matches.size() >= 4)
        {
            for( int i = 0; i < good_matches.size(); i++ )
            {
                //Get the keypoints from the good matches
                obj.push_back( kp_object[ good_matches[i].queryIdx ].pt );
                scene.push_back( kp_image[ good_matches[i].trainIdx ].pt );
           }

            H = findHomography( obj, scene, CV_RANSAC );

            perspectiveTransform( obj_corners, scene_corners, H);

            //Draw lines between the corners (the mapped object in the scene image )
            line( image, scene_corners[0] + Point2f( object.cols, 0), scene_corners[1] + Point2f( object.cols, 0), Scalar(0, 255, 0), 4 );
            line( image, scene_corners[1] + Point2f( object.cols, 0), scene_corners[2] + Point2f( object.cols, 0), Scalar( 0, 255, 0), 4 );
            line( image, scene_corners[2] + Point2f( object.cols, 0), scene_corners[3] + Point2f( object.cols, 0), Scalar( 0, 255, 0), 4 );
            line( image, scene_corners[3] + Point2f( object.cols, 0), scene_corners[0] + Point2f( object.cols, 0), Scalar( 0, 255, 0), 4 );
        }

        //Show detected matches
        imshow( "Good Matches", image );

        key = waitKey(1);
    }
    return 0;
}