Back to Search Start Over

CAESR: Conditional Autoencoder and Super-Resolution for Learned Spatial Scalability

Authors :
Bonnineau, Charles
Hamidouche, Wassim
Travers, Jean-François
Sidaty, Naty
Aubié, Jean-Yves
Deforges, Olivier
Source :
2021 International Conference on Visual Communications and Image Processing (VCIP)
Publication Year :
2022

Abstract

In this paper, we present CAESR, an hybrid learning-based coding approach for spatial scalability based on the versatile video coding (VVC) standard. Our framework considers a low-resolution signal encoded with VVC intra-mode as a base-layer (BL), and a deep conditional autoencoder with hyperprior (AE-HP) as an enhancement-layer (EL) model. The EL encoder takes as inputs both the upscaled BL reconstruction and the original image. Our approach relies on conditional coding that learns the optimal mixture of the source and the upscaled BL image, enabling better performance than residual coding. On the decoder side, a super-resolution (SR) module is used to recover high-resolution details and invert the conditional coding process. Experimental results have shown that our solution is competitive with the VVC full-resolution intra coding while being scalable.

Details

Database :
arXiv
Journal :
2021 International Conference on Visual Communications and Image Processing (VCIP)
Publication Type :
Report
Accession number :
edsarx.2202.00416
Document Type :
Working Paper
Full Text :
https://doi.org/10.1109/VCIP53242.2021.9675351