I am self learning image registration using matlab.
Suppose I want to study the shrinking speed of a water pool in a desert within a month. The camera's position and orientation is fixed on a tree nearby. Three markers are placed around the pool as a reference, like this:
then I took an image to record its initial position. But I forgot to record the elevation of the camera.
After a month I came back, I noticed that my camera has been stolen and the tree cut down, but the markers' inital positions still. I have to continue the study and then took another image that includes the pool and all the three initial markers, but this time at a different elevation and orientation, both of which are unknown.
Can I still accurately measure the ratio of pool size change? If yes, which transformation shall I apply to achieve this? Affine theorem or projective theorem?