Cell-based 2-step scalar deadzone quantization for high bit-depth hyperspectral image coding

Web of Science: 3 cites, Scopus: 5 cites, Google Scholar: cites

(Universitat Autònoma de Barcelona. Departament d'Enginyeria de la Informació i de les Comunicacions)
Aulí Llinàs, Francesc

(Universitat Autònoma de Barcelona. Departament d'Enginyeria de la Informació i de les Comunicacions)

Data:	2015
Resum:	Remote sensing images often need to be coded and/or transmitted with constrained computational resources. Among other features, such images commonly have high spatial, spectral, and bit-depth resolution, which may render difficult their handling. This letter introduces an embedded quantization scheme based on two-step scalar deadzone quantization (2SDQ) that enhances the quality of transmitted images when coded with a constrained number of bits. The proposed scheme is devised for use in JPEG2000. It is named cell-based 2SDQ since it uses cells, i. e. , small sets of wavelet coefficients within the codeblocks defined by JPEG2000. Cells permit a finer discrimination of coefficients in which to apply the proposed quantizer. Experimental results indicate that the proposed scheme is especially beneficial for high bit-depth hyperspectral images.
Ajuts:	Ministerio de Economía y Competitividad RYC-2010-05671 Ministerio de Economía y Competitividad TIN2012-38102-C03-03 Agència de Gestió d'Ajuts Universitaris i de Recerca 2014/SGR-691
Drets:	Tots els drets reservats.
Llengua:	Anglès
Document:	Article ; recerca ; Versió sotmesa a revisió
Matèria:	Embedded quantization ; High bit-depth images ; JPEG2000 ; Two-step scalar deadzone quantization (2SDQ)
Publicat a:	IEEE geoscience and remote sensing letter, Vol. 12 no. 9 (September 2015) p. 1893-1897

Pre-print
5 p, 974.5 KB

Registre creat el 2015-12-15, darrera modificació el 2022-09-04