Dual-lens super-resolution (DuSR) aims to improve a wide-angle low-resolution image by exploiting a telephoto reference image. A key difficulty is making reliable use of that reference when the two lenses have large parallax, which can cause misalignment and artifacts.
The paper proposes DuSRFlow, a method that combines parallax-aware flow estimation with complementary reference warping. This is designed to better handle the geometric differences between the wide-angle and telephoto views, allowing the reference to contribute useful detail without introducing errors.
The abstract is brief, but it clearly identifies large parallax as the central challenge and positions DuSRFlow as a solution. No experimental results are described in the available text, so the claims are limited to the method's design and intended purpose.