Oriented-grid Encoder for 3D Implicit Representations
This work improves 3D geometric encoding for object and scene representation, though it appears incremental as it builds on existing grid-based approaches.
The paper tackles the problem of encoding 3D points for implicit scene representation by proposing a 3D-oriented grid encoder with cylindrical volumetric interpolation, which exploits surface normals and local smoothness. The method converges faster and achieves sharper surfaces than regular grids, obtaining state-of-the-art results on datasets like ABC, Thingi10k, ShapeNet, and Matterport3D.
Encoding 3D points is one of the primary steps in learning-based implicit scene representation. Using features that gather information from neighbors with multi-resolution grids has proven to be the best geometric encoder for this task. However, prior techniques do not exploit some characteristics of most objects or scenes, such as surface normals and local smoothness. This paper is the first to exploit those 3D characteristics in 3D geometric encoders explicitly. In contrast to prior work on using multiple levels of details, regular cube grids, and trilinear interpolation, we propose 3D-oriented grids with a novel cylindrical volumetric interpolation for modeling local planar invariance. In addition, we explicitly include a local feature aggregation for feature regularization and smoothing of the cylindrical interpolation features. We evaluate our approach on ABC, Thingi10k, ShapeNet, and Matterport3D, for object and scene representation. Compared to the use of regular grids, our geometric encoder is shown to converge in fewer steps and obtain sharper 3D surfaces. When compared to the prior techniques, our method gets state-of-the-art results.