Methods to create montages of wide-field fundus images, while correcting for the projective distortion inherent to an imaging system are described. The methods use specific knowledge of the imaging geometry of the particular imaging system to map the fundus images onto a set of 3D ray vectors, which allows them to be stitched together efficiently and precisely. After co-aligning the images using feature detection and matching, the registered images are projected back into 2D to generate the final montaged image. The method could be used on any type of wide-field fundus images including, but not limited to, those generated from optical coherence tomography, optical coherence tomography angiography, and broad-line fundus imaging systems.
