Classification of Hyperspectral Images Compressed through 3D-JPEG2000

  • Authors:

  • Ian Blanes;Alaitz Zabala;Gerard Moré;Xavier Pons;Joan Serra-Sagristà

  • Affiliations:

  • Department of Information and Communications Engineering, ,;Department of Geography, ,;Centre for Ecological Research and Forestry Applications (CREAF), Universitat Autònoma de Barcelona, 08290, Spain;Department of Geography, , and Centre for Ecological Research and Forestry Applications (CREAF), Universitat Autònoma de Barcelona, 08290, Spain;Department of Information and Communications Engineering, ,

  • Venue:
  • KES '08 Proceedings of the 12th international conference on Knowledge-Based Intelligent Information and Engineering Systems, Part III
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

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.