Towards Practical Lossless Neural Compression for LiDAR Point Clouds

Abstract

LiDAR point clouds are fundamental to various applications, yet the extreme sparsity of high-precision geometric details hinders efficient context modeling, thereby limiting the compression speed and performance of existing methods. To address this challenge, we propose a compact representation for efficient predictive lossless coding. Our framework comprises two lightweight modules. First, the Geometry Re-Densification Module iteratively densifies encoded sparse geometry, extracts features at a dense scale, and then sparsifies the features for predictive coding. This module avoids costly computation on highly sparse details while maintaining a lightweight prediction head. Second, the Cross-scale Feature Propagation Module leverages occupancy cues from multiple resolution levels to guide hierarchical feature propagation, enabling information sharing across scales and reducing redundant feature extraction. Additionally, we introduce an integer-only inference pipeline to enable bit-exact cross-platform consistency, which avoids the entropy-coding collapse observed in existing neural compression methods and further accelerates coding. Experiments demonstrate competitive compression performance at real-time speed. Code will be released upon acceptance. Code is available at https://github.com/pengpeng-yu/FastPCC.

Figures (12)

• Figure 1: Illustration of the High-Resolution Contextual Sparsity (HRCS) phenomenon: (a) and (b) depict the voxelized octree representations at levels 8 and 12 for a point cloud from the KITTI dataset, respectively. The red bounding box highlights a $3 \times 3 \times 3$ neighborhood. (c) quantifies HRCS on the KITTI dataset, where the average number of neighbors per node decreases sharply with increasing octree level.
• Figure 2: (a) RD performance versus coding speed on KITTI at 12-bit precision, evaluated on an AMD EPYC 7R32 CPU and an NVIDIA RTX 4090 GPU. (b) Decoding failure caused by non-deterministic floating-point computation, where encoding is performed on an NVIDIA RTX 4090 GPU and decoding on an RTX 5880 GPU. The reconstructed point cloud collapses into an approximately uniform distribution in 3D space. (c) Bit-exact decoding achieved with the proposed integer-only inference pipeline.
• Figure 3: Pipeline of compressing a single octree level in the proposed lossless LiDAR PCC framework. The framework consists of four main stages: octree construction, prior construction, cross-scale feature propagation, and entropy coding. The cross-scale feature propagation module comprises two key components: the shallow-level propagation block and the deep-level propagation block, both adapted from the geometry re-densification module to exploit cross-scale features.
• Figure 4: RD performance comparison with existing methods on KITTI and Ford datasets from 11 bits to 16 bits.
• Figure 5: Comparison of coding time across 11--16 bits with existing real-time methods.