Octree-based Learned Point Cloud Geometry Compression: A Lossy Perspective

2026-03-30

AI Summary

This work presents the first systematic exploration of lossy point cloud compression within a learned octree framework, addressing the severe distortion and point loss commonly caused by quantization step adjustments in traditional octree-based methods. For object point clouds, the authors propose a leaf-node-based bitplane coding scheme coupled with binary occupancy prediction. For LiDAR point clouds, they introduce a fine-tuning-free variable-rate control strategy. By integrating contextual learning with efficient entropy coding, the proposed approach substantially outperforms existing octree-based methods on object point clouds, while achieving approximately 1% rate-control error on LiDAR data—demonstrating a strong balance between reconstruction fidelity and practical applicability.

Abstract

Octree-based context learning has recently become a leading method in point cloud compression. However, its potential on lossy compression remains undiscovered. The traditional lossy compression paradigm using lossless octree representation with quantization step adjustment may result in severe distortions due to massive missing points in quantization. Therefore, we analyze data characteristics of different point clouds and propose lossy approaches specifically. For object point clouds that suffer from quantization step adjustment, we propose a new leaf nodes lossy compression method, which achieves lossy compression by performing bit-wise coding and binary prediction on leaf nodes. For LiDAR point clouds, we explore variable rate approaches and propose a simple but effective rate control method. Experimental results demonstrate that the proposed leaf nodes lossy compression method significantly outperforms the previous octree-based method on object point clouds, and the proposed rate control method achieves about 1% bit error without finetuning on LiDAR point clouds.

Problem

• point cloud compression
• lossy compression
• octree
• quantization distortion
• rate control

Innovation

• octree-based compression
• lossy point cloud compression
• leaf node coding
• rate control
• binary prediction