OctHilNet: Hilbert-Guided Hierarchical Geometry Codec for Octree-Structured LiDAR Point Clouds
Mingjian Feng, Mingyue Cui, Yuyang Zhong, Chunjie Shu, Han Liu, Daosong Hu, Kai Huang
AI summary
Problem
Existing octree-based LiDAR point cloud compression methods fail to fully utilize spatial context due to uneven scanning density and the decoupling of sequential adjacency from geometric proximity in standard node ordering.
Approach
The method transforms points to polar coordinates with non-linear radial rebalancing to fix density imbalance, reorganizes octree nodes using a Hilbert space-filling curve to maintain geometric proximity, and employs a hierarchical Transformer with LocAtten and NeighbConv modules to capture fine-grained local dependencies.
Key results
- 45.1%-50.1% BD-Rate gain on SemanticKITTI dataset
- 51.9%-53.9% BD-Rate gain on Ford dataset
- Polar coordinate transformation with non-linear radial rebalancing for uniform point distribution
- Hilbert-guided sorting and LocAtten/NeighbConv modules for fine-grained spatial correlation capture
Why it matters
It enables more efficient storage and transmission of 3D LiDAR data while maintaining high reconstruction quality for downstream tasks like vehicle detection and semantic segmentation.
Abstract
High-quality LiDAR point cloud (LPC) compres- sion is essential for the storage and transmission of 3D data. The octree-structured entropy codec has emerged as the predom- inant method; however, previous methods do not fully utilize spatial contextual information, due to the loss of local features caused by uneven scanning density. To address this problem, we propose OctHilNet, a novel Hilbert-guided hierarchical framework for LPC compression that introduces the polarized octree for efficient node organization and the serialize-driven entropy model to strengthen the continuity of node contexts. Specifically, to counteract the inherent density imbalance, OctHilNet first transforms points into polar coordinates and applies a non-linear rebalancing to the radial distance. Then, we introduce the Hilbert space-filling curve to mitigate the impact of the decoupling between sequential adjacency and geometric proximity in octree node sequences. Finally, to better capture fine-grained spatial correlations, we propose LocAt- ten and NeighbConv modules in a hierarchical Transformer, which jointly strengthen local dependencies overlooked by standard self-attention. Compared to the previous state-of-the- art works, our method achieves 45.1%-50.1% and 51.9%-53.9% BD-Rate gains on the LPC benchmark SemanticKITTI and MPEG-specified Ford datasets, respectively. In particular, our OctHilNet allows for extension to downstream tasks (i.e., vehicle detection and semantic segmentation), further demonstrating the practicality of the method.