Suppose that I have two photos, I1 and I2 and I want to find a set of corresponding points between the two photos to define a homography between them. There are many algorithms for this purpose, namely SIFT which is scale invariant. The problem is that when there's a pattern in both photos (like repeated windows), SIFT starts to fail and most of the corresponding points it gives are false and outliers. So, what can I do in cases like these?
How can I avoid this problem? Are there other algorithms that behave better when there's a pattern in both photos?