Rate-Distortion Optimized Azimuth Quantization for MPEG L3C2 LiDAR Point Cloud Compression

As a core data format for representing 3D objects, point clouds have seen widespread adoption, driven by rapid progress in 3D sensing and rendering technologies. To enable efficient storage and transmission of large-scale data, MPEG developed the Low-Latency, Low-Complexity LiDAR Point Cloud Compression (L3C2) standard. Although L3C2 has significantly improved the compression efficiency of LiDAR point clouds, there remains potential for further optimization, particularly in distortion-sensitive azimuth information. A more sophisticated quantization strategy is therefore needed to strike an optimal balance between compression efficiency and the preservation of essential geometry details. This paper presents a rate-distortion optimized azimuth quantization method for MPEG L3C2. Specifically, we propose a bi-directional azimuth quantization strategy and achieve the adaptive selection of the optimal quantization method through rate-distortion optimization, thereby achieving a better balance between distortion and bitrate. Experimental evaluations verify that the proposed technique surpasses the MPEG L3C2 standard, delivering more effective LiDAR point cloud compression.