Improved Deep Point Cloud Geometry Compression

TL;DR

This paper introduces several novel techniques for deep point cloud compression, significantly improving rate-distortion performance over existing methods through hyperprior models, deeper transforms, and optimized training strategies.

Contribution

It proposes a comprehensive set of improvements including a scale hyperprior model, deeper transforms, and sequential training, with extensive ablation studies to understand their impact.

Findings

Achieves BD-PSNR gains of over 5 dB compared to G-PCC methods.

Demonstrates the effectiveness of each proposed component through ablation studies.

Provides open-source code for reproducibility.

Abstract

Point clouds have been recognized as a crucial data structure for 3D content and are essential in a number of applications such as virtual and mixed reality, autonomous driving, cultural heritage, etc. In this paper, we propose a set of contributions to improve deep point cloud compression, i.e.: using a scale hyperprior model for entropy coding; employing deeper transforms; a different balancing weight in the focal loss; optimal thresholding for decoding; and sequential model training. In addition, we present an extensive ablation study on the impact of each of these factors, in order to provide a better understanding about why they improve RD performance. An optimal combination of the proposed improvements achieves BD-PSNR gains over G-PCC trisoup and octree of 5.50 (6.48) dB and 6.84 (5.95) dB, respectively, when using the point-to-point (point-to-plane) metric. Code is available at https://github.com/mauriceqch/pcc_geo_cnn_v2 .