Semantic Communication for Efficient Point Cloud Transmission

TL;DR

The paper addresses efficient wireless transmission of dense 3D point clouds by introducing a semantic communication framework that separately extracts local and global semantics. It employs a patch-based local encoder and a projection-based global encoder, transmitting global information losslessly and local information via joint source-channel coding, under a two-stage training regime for rapid adaptation to channel conditions. The approach demonstrates superior reconstruction quality over traditional compression methods (e.g., G-PCC) and competing SemCom systems, achieving high PSNR even under severe noise, while maintaining a lightweight model suitable for edge devices. This method advances practical point cloud transmission by combining multi-level semantics with lossless-lossy layering and efficient training, enabling robust, bandwidth-efficient recovery in wireless scenarios.

Abstract

As three-dimensional acquisition technologies like LiDAR cameras advance, the need for efficient transmission of 3D point clouds is becoming increasingly important. In this paper, we present a novel semantic communication (SemCom) approach for efficient 3D point cloud transmission. Different from existing methods that rely on downsampling and feature extraction for compression, our approach utilizes a parallel structure to separately extract both global and local information from point clouds. This system is composed of five key components: local semantic encoder, global semantic encoder, channel encoder, channel decoder, and semantic decoder. Our numerical results indicate that this approach surpasses both the traditional Octree compression methodology and alternative deep learning-based strategies in terms of reconstruction quality. Moreover, our system is capable of achieving high-quality point cloud reconstruction under adverse channel conditions, specifically maintaining a reconstruction quality of over 37dB even with severe channel noise.

Figures (6)

• Figure 1: The overall framework of our point cloud semantic transmission system
• Figure 2: The specific structure of each module, from left to right, is the local semantic encoder, global semantic encoder, channel encoder and decoder, semantic decoder.
• Figure 3: The reconstruction performance of the model under different compression performance, we set SNR=$\{10,5,4\}$. At the same time, we show the performance of SEPT when bottleneck size=$300$.
• Figure 4: Comparison between the reconstruction performance of our model and DPCC under different SNR.
• Figure 5: Visualization results of the three models.