Finding right point matches using RANSAC

Question

I have 2 images. Each has 8 points from the corners of two squares in each image. I need to find the corresponding correct matches between them. So, when I pass the 64 possible combinations of matches into RANSAC (findFundamentalMatrix), it gave me 7 matches which were suppose to be inliers but, they turned out to be all wrong matches.

I have been scratching my head since many days trying to figure out what could be wrong. Since I am new to OpenCV, I need some help figuring it out.

Thanks

Answer 1

For solve this problem in the first use findHomography() then compute rotation value in the last use perspectiveTransform() to transform input points to original points.

  Point2f first_8_points[8],second_8_points[8];
  Mat H = findHomography(first_8_points, second_8_points, CV_RANSAC );
  float roatation =  acos( H.at<double>(0,0)) * 180/CV_PI;
  perspectiveTransform( first_8_points, second_8_points, H);