How to transform new points after the matches were found?

[thiagoribeirodamotta]

Hi there,

First of all, I'd like to thanks for the great contribution! I'm looking forward to keep on trying this out!

The scenario I'm currently working on requires the transformation matrix that takes points placed on Image1 and correctly place them on Image2.

Since I couldn't find where in the code this transformation happens, I just followed the demo.py script and added the following at the end of it:

  src_pts = np.float32(match_kp0).reshape(-1,1,2)
  dst_pts = np.float32(match_kp1).reshape(-1,1,2)
  M, _ = cv.findHomography(src_pts, dst_pts, cv.RANSAC,3.0)

  transformed_points_to_image_2= cv.perspectiveTransform(np.float32(new_points_image_1).reshape(-1,1,2), M)

Would you say this is the correct approach to transform new points from image 1 coordinate space to image 2 coordinate space?