Multi-Class Boundary Extraction from Implicit Representations
This work addresses a gap in multi-class surface extraction for applications like geological modelling, but it is incremental as it extends existing single-class methods to the multi-class case.
The paper tackles the problem of extracting boundaries from implicit neural representations for multiple classes, which lacked methods ensuring topological correctness and no holes, by introducing a 2D algorithm that guarantees these properties and allows for minimum detail restraint, evaluated on geological modelling data to show adaptiveness and handling of complex topology.
Surface extraction from implicit neural representations modelling a single class surface is a well-known task. However, there exist no surface extraction methods from an implicit representation of multiple classes that guarantee topological correctness and no holes. In this work, we lay the groundwork by introducing a 2D boundary extraction algorithm for the multi-class case focusing on topological consistency and water-tightness, which also allows for setting minimum detail restraint on the approximation. Finally, we evaluate our algorithm using geological modelling data, showcasing its adaptiveness and ability to honour complex topology.