NEF: Neural Edge Fields for 3D Parametric Curve Reconstruction from Multi-view Images

TL;DR

This paper introduces Neural Edge Field (NEF), a neural implicit representation for reconstructing 3D feature curves from multi-view images, leveraging view-based rendering loss without requiring explicit 3D edge supervision.

Contribution

The paper proposes NEF, a novel neural implicit model that reconstructs 3D edges directly from multi-view images using a rendering-based loss, eliminating the need for 3D edge annotations or correspondences.

Findings

NEF outperforms state-of-the-art methods on synthetic data benchmarks.

The approach effectively exploits 2D edge detection without 3D supervision.

The method produces robust and accurate 3D curve reconstructions.

Abstract

We study the problem of reconstructing 3D feature curves of an object from a set of calibrated multi-view images. To do so, we learn a neural implicit field representing the density distribution of 3D edges which we refer to as Neural Edge Field (NEF). Inspired by NeRF, NEF is optimized with a view-based rendering loss where a 2D edge map is rendered at a given view and is compared to the ground-truth edge map extracted from the image of that view. The rendering-based differentiable optimization of NEF fully exploits 2D edge detection, without needing a supervision of 3D edges, a 3D geometric operator or cross-view edge correspondence. Several technical designs are devised to ensure learning a range-limited and view-independent NEF for robust edge extraction. The final parametric 3D curves are extracted from NEF with an iterative optimization method. On our benchmark with synthetic data, we demonstrate that NEF outperforms existing state-of-the-art methods on all metrics. Project page: https://yunfan1202.github.io/NEF/.