Regression Wavelet Analysis for Near-Lossless Remote Sensing Data Compression

Regression wavelet analysis (RWA) is one of the current state-of-the-art lossless compression techniques for remote sensing data. This article presents the first regression-based near-lossless compression method. It is built upon RWA, a quantizer, and a feedback loop to compensate the quantization e...

Descripción completa

Detalles Bibliográficos
Autores: Álvarez Cortés, Sara|||0000-0003-3244-8080, Serra-Sagristà, Joan|||0000-0003-4729-9292, Bartrina-Rapesta, Joan|||0000-0002-1551-3680, Marcellin, Michael W.|||0000-0001-9606-134X
Tipo de recurso: artículo
Fecha de publicación:2020
País:España
Institución:Universitat Autònoma de Barcelona
Repositorio:Dipòsit Digital de Documents de la UAB
Idioma:inglés
OAI Identifier:oai:ddd.uab.cat:215773
Acceso en línea:https://ddd.uab.cat/record/215773
https://dx.doi.org/urn:doi:10.1109/TGRS.2019.2940553
Access Level:acceso abierto
Palabra clave:Lossless and near-lossless compression
Pyramidal multiresolution scheme
Regression wavelet analysis
Remote sensing data compression
Descripción
Sumario:Regression wavelet analysis (RWA) is one of the current state-of-the-art lossless compression techniques for remote sensing data. This article presents the first regression-based near-lossless compression method. It is built upon RWA, a quantizer, and a feedback loop to compensate the quantization error. Our near-lossless RWA (NLRWA) proposal can be followed by any entropy coding technique. Here, the NLRWA is coupled with a bitplane-based coder that supports progressive decoding. This successfully enables gradual quality refinement and lossless and near-lossless recovery. A smart strategy for selecting the NLRWA quantization steps is also included. Experimental results show that the proposed scheme outperforms the state-of-the-art lossless and the near-lossless compression methods in terms of compression ratios and quality retrieval.