Double-Deep Learning-Based Point Cloud Geometry Coding with Adaptive Super-Resolution

Ruivo, Manuel; Guarda, Andre F. R.; Pereira, Fernando

doi:10.1109/euvip53989.2022.9922784

Cited by 4 publications

(2 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Instead, using bigger block sizes during compression or higher values for the quantization step could become viable alternatives. Moreover, compressing downsampled versions of the point cloud and performing upsampling as post-processing, similarly to the method proposed by Ruivo et al, 47 could be another way of achieving higher compression rates for both geometry and color.…”

Section: Objective Resultsmentioning

confidence: 99%

“…Future work could aim at allowing the integrated codec to achieve lower bitrates and further improve its rate-distortion performance, specially for color coding. For the first task, applying downsampling prior to compression and upsampling as a post-processing step, similarly to what is proposed by Ruivo et al 47 for the geometric coordinates, could be a viable solution. Since the most part of the bitstream is associated with color information, an adaptation of the previous method to our color coding method could be envisaged.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Integrated learning-based point cloud compression for geometry and color with graph Fourier transforms

Lazzarotto

Ebrahimi

2022

Applications of Digital Image Processing XLV

View full text Add to dashboard Cite

Point cloud representation is a popular modality to code immersive 3D contents. Several solutions and standards have been recently proposed in order to efficiently compress the large volume of data that point clouds require, in order to make them feasible for real-life applications. Recent studies adopting learning-based methods for point cloud compression have demonstrated high compression efficiency specially when compared to the conventional compression standards. However, they are mostly evaluated either on geometry or color separately, and few learning-based joint codecs have been proposed. In this paper, we propose an integrated learned coding architecture by joining a previously proposed geometry coding module based on three-dimensional convolutional layers with a color compression method relying on graph Fourier transform (GFT) using a learning-based mean and scale hyperprior to compress the obtained coefficients. Evaluation on a test set with dense point clouds shows that the proposed method outperforms G-PCC and achieves competitive performance with V-PCC when evaluated with state-of-the-art objective quality metrics.

show abstract

Section: Objective Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%