Serialization based Point Cloud Oversegmentation

摘要

Point cloud oversegmentation, as a fundamental preprocessing step for 3D understanding, is a challenging task due to its spatial proximity and semantic similarity requirements. Most existing works struggle to efficiently group semantically consistent points into superpoints while maintaining spatial proximity. In this paper, we propose a novel serialization based point cloud oversegmentation method, which leverages serialization to avoid complex spatial queries, directly accessing neighboring points through sequence locality for similarity matching and superpoint clustering. Specifically, we first serialize point clouds into a Hilbert curve and spatially-continuously partition them into initial segments. Then, to guarantee the internal semantic consistency of superpoints, we design an adaptive update algorithm that clusters superpoints by matching feature similarities between neighboring segments and refines segment features via Cross-Attention. Experiments on largescale indoor and outdoor datasets demonstrate state-of-the-art performance in point cloud oversegmentation. Moreover, it is also adaptable to semantic segmentation and achieves promising performance. The code is available at https://github.com/CHL-glitch/SPCNet.