One paper accepted by IEEE MIPR!

Written by  |  14/05/2022 - 09/02

Congrats to Jianqiang!

Title: Sparse Tensor-based Point Cloud Attribute Compression


Recently, numerous learning-based compression methods have been developed with outstanding performance for the compression of geometry information of point clouds.  On the contrary, limited explorations have been devoted to point cloud attribute compression (PCAC). Thus, this study focuses on the PCAC by applying sparse convolution because of its superior efficiency for representing the geometry of unstructured points. The proposed method stacks sparse convolutions to construct the variational autoencoder (VAE) framework to  compress the color attributes of a given point cloud. To better encode latent elements at the bottleneck, we apply the adaptive entropy model with the joint utilization of hyper prior and autoregressive neighbors to accurately estimate the bit rate.  The qualitative measurement of the proposed method already rivals  the latest G-PCC (or TMC13) version 14 at a similar bit rate. And,  our method shows clear quantitative improvements to G-PCC version 6, and largely outperforms existing learning-based methods, which promises encouraging potentials for learnt PCAC.