Multilevel split regression wavelet analysis for lossless compression of remote sensing data

Spectral redundancy is a key element to be exploited in compression of remote sensing data. Combined with an entropy encoder, it can achieve competitive lossless coding performance. One of the latest techniques to decorrelate the spectral signal is the regression wavelet analysis (RWA). RWA applies...

Descripción completa

Detalles Bibliográficos
Autores: Álvarez Cortés, Sara|||0000-0003-3244-8080, Bartrina-Rapesta, Joan|||0000-0002-1551-3680, Serra-Sagristà, Joan|||0000-0003-4729-9292
Tipo de recurso: artículo
Fecha de publicación:2018
País:España
Institución:Universitat Autònoma de Barcelona
Repositorio:Dipòsit Digital de Documents de la UAB
Idioma:inglés
OAI Identifier:oai:ddd.uab.cat:200081
Acceso en línea:https://ddd.uab.cat/record/200081
https://dx.doi.org/urn:doi:10.1109/LGRS.2018.2850938
Access Level:acceso abierto
Palabra clave:Lossless coding
Predictive coding
Spectral decorrelation
Descripción
Sumario:Spectral redundancy is a key element to be exploited in compression of remote sensing data. Combined with an entropy encoder, it can achieve competitive lossless coding performance. One of the latest techniques to decorrelate the spectral signal is the regression wavelet analysis (RWA). RWA applies a wavelet transform in the spectral domain and estimates the detail coeffi- cients through the approximation coefficients using linear regres- sion. RWA was originally coupled with JPEG 2000. This letter introduces a novel coding approach, where RWA is coupled with the predictor of CCSDS-123.0-B-1 standard and a lightweight contextual arithmetic coder. In addition, we also propose a smart strategy to select the number of RWA decomposition levels that maximize the coding performance. Experimental results indicate that, on average, the obtained coding gains vary between 0.1 and 1.35 bits-per-pixel-per-component compared with the other state- of-the-art coding techniques.