ANFIC: Image Compression Using Augmented Normalizing Flows

This paper introduces an end-to-end learned image compression system, termed ANFIC, based on Augmented Normalizing Flows (ANF). ANF is a new type of flow model, which stacks multiple variational autoencoders (VAE) for greater model expressiveness. The VAE-based image compression has gone mainstream,...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Yung-Han Ho, Chih-Chun Chan, Wen-Hsiao Peng, Hsueh-Ming Hang, Marek Domanski
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2021
Materias:	Learning-based image compression flow-based image compression augmented normalizing flows perceptually lossless image compression variable rate image compression Electric apparatus and materials. Electric circuits. Electric networks TK452-454.4
Acceso en línea:	https://doaj.org/article/c929fe6d235b4cfdaefe91c5b076a3cc
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:c929fe6d235b4cfdaefe91c5b076a3cc
record_format	dspace
spelling	oai:doaj.org-article:c929fe6d235b4cfdaefe91c5b076a3cc2021-11-23T00:02:16ZANFIC: Image Compression Using Augmented Normalizing Flows2644-122510.1109/OJCAS.2021.3123201https://doaj.org/article/c929fe6d235b4cfdaefe91c5b076a3cc2021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9623351/https://doaj.org/toc/2644-1225This paper introduces an end-to-end learned image compression system, termed ANFIC, based on Augmented Normalizing Flows (ANF). ANF is a new type of flow model, which stacks multiple variational autoencoders (VAE) for greater model expressiveness. The VAE-based image compression has gone mainstream, showing promising compression performance. Our work presents the first attempt to leverage VAE-based compression in a flow-based framework. ANFIC advances further compression efficiency by stacking and extending hierarchically multiple VAE’s. The invertibility of ANF, together with our training strategies, enables ANFIC to support a wide range of quality levels without changing the encoding and decoding networks. Extensive experimental results show that in terms of PSNR-RGB, ANFIC performs comparably to or better than the state-of-the-art learned image compression. Moreover, it performs close to VVC intra coding, from low-rate compression up to perceptually lossless compression. In particular, ANFIC achieves the state-of-the-art performance, when extended with conditional convolution for variable rate compression with a single model. The source code of ANFIC can be found at <uri>https://github.com/dororojames/ANFIC</uri>.Yung-Han HoChih-Chun ChanWen-Hsiao PengHsueh-Ming HangMarek DomanskiIEEEarticleLearning-based image compressionflow-based image compressionaugmented normalizing flowsperceptually lossless image compressionvariable rate image compressionElectric apparatus and materials. Electric circuits. Electric networksTK452-454.4ENIEEE Open Journal of Circuits and Systems, Vol 2, Pp 613-626 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Learning-based image compression flow-based image compression augmented normalizing flows perceptually lossless image compression variable rate image compression Electric apparatus and materials. Electric circuits. Electric networks TK452-454.4
spellingShingle	Learning-based image compression flow-based image compression augmented normalizing flows perceptually lossless image compression variable rate image compression Electric apparatus and materials. Electric circuits. Electric networks TK452-454.4 Yung-Han Ho Chih-Chun Chan Wen-Hsiao Peng Hsueh-Ming Hang Marek Domanski ANFIC: Image Compression Using Augmented Normalizing Flows
description	This paper introduces an end-to-end learned image compression system, termed ANFIC, based on Augmented Normalizing Flows (ANF). ANF is a new type of flow model, which stacks multiple variational autoencoders (VAE) for greater model expressiveness. The VAE-based image compression has gone mainstream, showing promising compression performance. Our work presents the first attempt to leverage VAE-based compression in a flow-based framework. ANFIC advances further compression efficiency by stacking and extending hierarchically multiple VAE’s. The invertibility of ANF, together with our training strategies, enables ANFIC to support a wide range of quality levels without changing the encoding and decoding networks. Extensive experimental results show that in terms of PSNR-RGB, ANFIC performs comparably to or better than the state-of-the-art learned image compression. Moreover, it performs close to VVC intra coding, from low-rate compression up to perceptually lossless compression. In particular, ANFIC achieves the state-of-the-art performance, when extended with conditional convolution for variable rate compression with a single model. The source code of ANFIC can be found at <uri>https://github.com/dororojames/ANFIC</uri>.
format	article
author	Yung-Han Ho Chih-Chun Chan Wen-Hsiao Peng Hsueh-Ming Hang Marek Domanski
author_facet	Yung-Han Ho Chih-Chun Chan Wen-Hsiao Peng Hsueh-Ming Hang Marek Domanski
author_sort	Yung-Han Ho
title	ANFIC: Image Compression Using Augmented Normalizing Flows
title_short	ANFIC: Image Compression Using Augmented Normalizing Flows
title_full	ANFIC: Image Compression Using Augmented Normalizing Flows
title_fullStr	ANFIC: Image Compression Using Augmented Normalizing Flows
title_full_unstemmed	ANFIC: Image Compression Using Augmented Normalizing Flows
title_sort	anfic: image compression using augmented normalizing flows
publisher	IEEE
publishDate	2021
url	https://doaj.org/article/c929fe6d235b4cfdaefe91c5b076a3cc
work_keys_str_mv	AT yunghanho anficimagecompressionusingaugmentednormalizingflows AT chihchunchan anficimagecompressionusingaugmentednormalizingflows AT wenhsiaopeng anficimagecompressionusingaugmentednormalizingflows AT hsuehminghang anficimagecompressionusingaugmentednormalizingflows AT marekdomanski anficimagecompressionusingaugmentednormalizingflows
_version_	1718417403406188544

ANFIC: Image Compression Using Augmented Normalizing Flows

Ejemplares similares