Initial bits generation for lossless image compression using bits-back coding with a latent variable model

Haruka Sugiyama; Kaiyu Suzuki; Ichiro Matsuda

doi:10.1117/12.3019366

2 May 2024 Initial bits generation for lossless image compression using bits-back coding with a latent variable model

Haruka Sugiyama, Kaiyu Suzuki, Ichiro Matsuda

Proceedings Volume 13164, International Workshop on Advanced Imaging Technology (IWAIT) 2024; 131642P (2024) https://doi.org/10.1117/12.3019366
Event: International Workshop on Advanced Imaging Technology (IWAIT) 2024, 2024, Langkawi, Malaysia

Abstract

Bits-back coding is a kind of entropy coding algorithm and can be used with an image generative model using latent variables for lossless image compression. It achieves excellent coding performance when ignoring the amount of initial bits which must be prepared independent of image contents. However, the initial bits are considered as additional information and indispensable for the encoding process. In this paper, a way of initial bits generation as well as discretization precision of the latent variables are investigated for better coding performance in a practical sense.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Haruka Sugiyama, Kaiyu Suzuki, and Ichiro Matsuda "Initial bits generation for lossless image compression using bits-back coding with a latent variable model", Proc. SPIE 13164, International Workshop on Advanced Imaging Technology (IWAIT) 2024, 131642P (2 May 2024); https://doi.org/10.1117/12.3019366

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available