Classification of hyperspectral images compressed through 3D-JPEG2000

Ian Blanes*, Alaitz Zabala, Gerard Moré, Xavier Pons, Joan Serra-Sagristà

*Corresponding author for this work

Research output: Contribution to journalArticleResearchpeer-review

11 Citations (Scopus)

Abstract

Classification of hyperspectral images is paramount to an increasing number of user applications. With the advent of more powerful technology, sensed images demand for larger requirements in computational and memory capabilities, which has led to devise compression techniques to alleviate the transmission and storage necessities. Classification of compressed images is addressed in this paper. Compression takes into account the spectral correlation of hyperspectral images together with more simple approaches. Experiments have been performed on a large hyperspectral CASI image with 72 bands. Both coding and classification results indicate that the performance of 3d-DWT is superior to the other two lossy coding approaches, providing consistent improvements of more than 10 dB for the coding process, and maintaining both the global accuracy and the percentage of classified area for the classification process.

Original languageAmerican English
Pages (from-to)416-423
Number of pages8
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Issue numberPART 3
DOIs
Publication statusPublished - 2008

Keywords

  • 3-dimensional coding
  • Classification
  • Hyperspectral images
  • JPEG2000 standard

Fingerprint Dive into the research topics of 'Classification of hyperspectral images compressed through 3D-JPEG2000'. Together they form a unique fingerprint.

Cite this