Human-centric panoramic imaging stitching

Abstract

We introduce a novel image mosaicing algorithm to generate 360° landscape images while also taking into account the presence of people at the boundaries between stitched images. Current image mosaicing techniques tend to fail when there is extreme parallax caused by nearby objects or moving objects at the boundary between images. This parallax causes ghosting or unnatural discontinuities in the image. To address this problem, we present an image mosaicing algorithm that is robust to parallax and misalignment, and is also able to preserve the important human-centric content, specifically faces. In particular, we find an optimal path between the boundary of two images that preserves color continuity and peoples' faces in the scene. Preliminary results show promising results of preserving close-up faces with parallax while also being able to generate a perceptually plausible 360° panoramic image.