NeUDF: Leaning Neural Unsigned Distance Fields with Volume Rendering
This addresses the problem of reconstructing real-world objects with open surfaces for computer vision and graphics applications, representing a novel method for a known bottleneck.
The paper tackles the limitation of existing neural implicit surface rendering methods, which are restricted to closed surfaces, by introducing NeUDF, a framework that reconstructs surfaces with arbitrary topologies from multi-view supervision, achieving significant outperformance over state-of-the-art methods on datasets like DTU, MGN, and Deep Fashion 3D.
Multi-view shape reconstruction has achieved impressive progresses thanks to the latest advances in neural implicit surface rendering. However, existing methods based on signed distance function (SDF) are limited to closed surfaces, failing to reconstruct a wide range of real-world objects that contain open-surface structures. In this work, we introduce a new neural rendering framework, coded NeUDF, that can reconstruct surfaces with arbitrary topologies solely from multi-view supervision. To gain the flexibility of representing arbitrary surfaces, NeUDF leverages the unsigned distance function (UDF) as surface representation. While a naive extension of an SDF-based neural renderer cannot scale to UDF, we propose two new formulations of weight function specially tailored for UDF-based volume rendering. Furthermore, to cope with open surface rendering, where the in/out test is no longer valid, we present a dedicated normal regularization strategy to resolve the surface orientation ambiguity. We extensively evaluate our method over a number of challenging datasets, including DTU}, MGN, and Deep Fashion 3D. Experimental results demonstrate that nEudf can significantly outperform the state-of-the-art method in the task of multi-view surface reconstruction, especially for complex shapes with open boundaries.