Generalized Nested Latent Variable Models For Lossy Coding Applied To Wind Turbine Scenarios

Rate-distortion optimization through neural networks has accomplished competitive results in compression efficiency and image quality. This learning-based approach seeks to minimize the compromise between compression rate and reconstructed image quality by automatically extracting and retaining cruc...

ver descrição completa

Detalhes bibliográficos
Autores: Pérez-Gonzalo, Raül, Espersen, Andreas, Agudo Martínez, Antonio
Formato: artículo
Estado:Versión aceptada para publicación
Fecha de publicación:2024
País:España
Recursos:Consejo Superior de Investigaciones Científicas (CSIC)
Repositorio:DIGITAL.CSIC. Repositorio Institucional del CSIC
OAI Identifier:oai:digital.csic.es:10261/387908
Acesso em linha:http://hdl.handle.net/10261/387908
https://api.elsevier.com/content/abstract/scopus_id/85216844645
Access Level:acceso abierto
Palavra-chave:Blade Inspections
Image Compression
Nested Models
Rate-distortion Loss
Wind Turbine
Descrição
Resumo:Rate-distortion optimization through neural networks has accomplished competitive results in compression efficiency and image quality. This learning-based approach seeks to minimize the compromise between compression rate and reconstructed image quality by automatically extracting and retaining crucial information, while discarding less critical details. A successful technique consists in introducing a deep hyperprior that operates within a 2-level nested latent variable model, enhancing compression by capturing complex data dependencies. This paper extends this concept by designing a generalized L-level nested generative model with a Markov chain structure. We demonstrate as L increases that a trainable prior is detrimental and explore a common dimensionality along the distinct latent variables to boost compression performance. As this structured framework can represent autoregressive coders, we outperform the hyperprior model and achieve state-of-the-art performance while reducing substantially the computational cost. Our experimental evaluation is performed on wind turbine scenarios to study its application on visual inspections.