A deep perceptual metric for 3D point clouds

TL;DR

This paper introduces a new perceptual loss function for 3D point cloud compression that better aligns with human perception, along with a novel voxel grid representation improving perceptual quality prediction.

Contribution

It proposes a perceptual loss function tailored for 3D point clouds and a truncated distance field voxel grid representation, enhancing perceptual quality assessment.

Findings

The proposed perceptual loss outperforms existing loss functions on subjective datasets.

Truncated distance field voxel grids lead to sparser latent spaces and better perceptual correlation.

Common loss functions like focal loss and weighted binary cross entropy poorly correlate with human perception.

Abstract

Point clouds are essential for storage and transmission of 3D content. As they can entail significant volumes of data, point cloud compression is crucial for practical usage. Recently, point cloud geometry compression approaches based on deep neural networks have been explored. In this paper, we evaluate the ability to predict perceptual quality of typical voxel-based loss functions employed to train these networks. We find that the commonly used focal loss and weighted binary cross entropy are poorly correlated with human perception. We thus propose a perceptual loss function for 3D point clouds which outperforms existing loss functions on the ICIP2020 subjective dataset. In addition, we propose a novel truncated distance field voxel grid representation and find that it leads to sparser latent spaces and loss functions that are more correlated with perceived visual quality compared to a binary representation. The source code is available at https://github.com/mauriceqch/2021_pc_perceptual_loss.