Mapping the Right Eye (bottom frame) into the perspective of the Left Eye (top frame) using a $3 \times 3$ homography matrix.
This projective transformation is mathematically defined as:
x' = (h00x + h01y + h02) / w
y' = (h10x + h11y + h12) / w
w = (h20x + h21y + 1.0)
Alignment is critical for Reference-Based Super-Resolution, allowing the network to sample pixel-perfect data from the secondary camera.