c++opencvcomputer-visiontransformationimage-stabilization

opencv estimateRigidTransform: How to get global scale?


I'm using openCV's estimateRigidTransform to stabilize a series of video frames, which works very well.

The function returns a 2x3 transformation matrix M:

|  a_11 a_12 b_1 |
| -a_12 a_11 b_2 |

As far as I understand the b elements represent the translation, and the a elements are used to do the scaling and rotation.

What I would like to do, is get global scale from M - a value that represents the camera zooming factor. How can I do calculate this correctly from a values?


To clarify a little more what I want to get, here is an example. Let's say estimateRigidTransform calculates a transformation matrix M for the second frame below regarding to the first frame:

The global zooming factor in this case should be aroud 2. My question is how to get this value from M?


Solution

  • Ok, after additional search I found a great answer here: https://math.stackexchange.com/questions/13150/extracting-rotation-scale-values-from-2d-transformation-matrix

    Where the "zooming" is represented by s_x and s_y. These two values will mostly be equal (since zoom in video captured by cameras is equal in both directions), so I can take either one.