When taking multiple photos, e.g., of a landscape, then how do we stich them together as fast as we can?
GPUs! Well ok, algorithms that can be implemented on GPUs that extract image features from all the images, match features across images, and calculate a homography. To make things look seemless, we also implemented a warping step, as well as blending.
I focused on homography extraction with RANSAC and the warping step, as well as the pipeline architecture of the project.
C++, Feature Detection & Matching, Homography, RANSAC, Warping, Blending, GPU, Stiching