Batch image processing. Transformation in key areas. Solution example?

Question

Image processing

Good day.
Please advise how to approach the solution of the problem - there are several hundred similar, but not completely the same type of photographs in the geometric sense. Each of them contains the same object - a la a palaroid frame, but on each the position or perspective of this frame is slightly changed. The task is the image in Figure 4.
I understand correctly that the script should work in 2 steps:
1 Analysis of the photo by key areas, for example, those that I highlighted in one photo with green and red circles.
2 Image modification: change perspective, rotate, crop.
Which way to think? What libraries, language to use? Any solution will do, be it a WIndows GUI or an example Python library.

Answer 1

In OpenCV you can.

Answer 2

Gonzalez and Woods to the rescue!
And if you are too lazy to implement a smart, but highly specialized thing yourself, understanding Hough transforms, all sorts of Gaussian Laplacians, matrix operations, etc., then you can take some ready-made fat library (like OpenCV, but this is a terrible overhead for such a task).

Answer 3

The task is not entirely clear, you need to "insert" a frame into the photo, into which to insert a piece of the desired photo, what kind of perversion is this over photographs?
Even if we exclude the meaninglessness of the task, it comes down to putting a layer on the picture, why are there transformations at all. Well, in general, exactly the same thing, as far as I remember, Picasa knows how to do it in its collages, so you can just put it in there and load the pictures there and make fun of it as you please.