Zero NeRF: Registration with Zero Overlap
This addresses a critical challenge in 3D reconstruction and registration for applications like robotics or augmented reality, offering a novel solution for scenarios with zero overlap.
The paper tackles the problem of aligning scene representations with minimal or no visual overlap by introducing Zero-NeRF, a projective surface registration method that enforces consistency between visible surfaces to constrain occluded geometry, achieving registration in real-world scenes with infinitesimal overlaps where prior methods fail.
We present Zero-NeRF, a projective surface registration method that, to the best of our knowledge, offers the first general solution capable of alignment between scene representations with minimal or zero visual correspondence. To do this, we enforce consistency between visible surfaces of partial and complete reconstructions, which allows us to constrain occluded geometry. We use a NeRF as our surface representation and the NeRF rendering pipeline to perform this alignment. To demonstrate the efficacy of our method, we register real-world scenes from opposite sides with infinitesimal overlaps that cannot be accurately registered using prior methods, and we compare these results against widely used registration methods.