3DAC: Learning Attribute Compression for Point Clouds

TL;DR

This paper introduces 3DAC, a deep learning-based attribute compression method for large-scale 3D point clouds that improves storage efficiency and reconstruction quality by modeling attribute transform coefficients with a deep entropy model.

Contribution

The paper presents a novel deep compression network, 3DAC, specifically designed for attribute compression in large-scale 3D point clouds, integrating transform and entropy modeling.

Findings

Achieves superior compression rates on ScanNet and SemanticKITTI datasets.

Provides high-quality reconstruction with reduced storage requirements.

Demonstrates effectiveness across indoor and outdoor large-scale point clouds.

Abstract

We study the problem of attribute compression for large-scale unstructured 3D point clouds. Through an in-depth exploration of the relationships between different encoding steps and different attribute channels, we introduce a deep compression network, termed 3DAC, to explicitly compress the attributes of 3D point clouds and reduce storage usage in this paper. Specifically, the point cloud attributes such as color and reflectance are firstly converted to transform coefficients. We then propose a deep entropy model to model the probabilities of these coefficients by considering information hidden in attribute transforms and previous encoded attributes. Finally, the estimated probabilities are used to further compress these transform coefficients to a final attributes bitstream. Extensive experiments conducted on both indoor and outdoor large-scale open point cloud datasets, including ScanNet and SemanticKITTI, demonstrated the superior compression rates and reconstruction quality of the proposed 3DAC.