Segmentation and classification of hyperspectral images using watershed transformation

Y. Tarabalka*, J. Chanussot, J. A. Benediktsson

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

451 Citations (Scopus)

Abstract

Hyperspectral imaging, which records a detailed spectrum of light for each pixel, provides an invaluable source of information regarding the physical nature of the different materials, leading to the potential of a more accurate classification. However, high dimensionality of hyperspectral data, usually coupled with limited reference data available, limits the performances of supervised classification techniques. The commonly used pixel-wise classification lacks information about spatial structures of the image. In order to increase classification performances, integration of spatial information into the classification process is needed. In this paper, we propose to extend the watershed segmentation algorithm for hyperspectral images, in order to define information about spatial structures. In particular, several approaches to compute a one-band gradient function from hyperspectral images are proposed and investigated. The accuracy of the watershed algorithms is demonstrated by the further incorporation of the segmentation maps into a classifier. A new spectral-spatial classification scheme for hyperspectral images is proposed, based on the pixel-wise Support Vector Machines classification, followed by majority voting within the watershed regions. Experimental segmentation and classification results are presented on two hyperspectral images. It is shown in experiments that when the number of spectral bands increases, the feature extraction and the use of multidimensional gradients appear to be preferable to the use of vectorial gradients. The integration of the spatial information from the watershed segmentation in the hyperspectral image classifier improves the classification accuracies and provides classification maps with more homogeneous regions, compared to pixel-wise classification and previously proposed spectral-spatial classification techniques. The developed method is especially suitable for classifying images with large spatial structures.

Original languageEnglish
Pages (from-to)2367-2379
Number of pages13
JournalPattern Recognition
Volume43
Issue number7
DOIs
Publication statusPublished - Jul 2010

Bibliographical note

Funding Information:
This research is supported in part by the Marie Curie Research Training Network “HYPER-I-NET”. The authors would like to thank Paolo Gamba and David Landgrebe for providing the hyperspectral data and James Tilton for providing the NASA Goddard's RHSEG software.

Funding Information:
About the Author —YULIYA TARABALKA received her Bachelor degree in Computer Science (2005) from the Ternopil Ivan Pul’uj State Technical University, Ukraine and M.Sc. degree in Signal and Image Processing (2007) from the Grenoble Institute of Technology (INPG), France. From July 2007 to January 2008 she worked as a Researcher with the Norwegian Defence Research Establishment, Norway. Since 2007, she is a Ph.D. student, pursuing a co-joint degree between University of Iceland, Iceland and INPG, France. Her current research work is funded by the “HYPER-I-NET” Marie Curie Research Training Network. Her research interests are in the areas of image processing, pattern recognition, hyperspectral imaging and development of efficient algorithms.

Other keywords

  • Classification
  • Hyperspectral images
  • Mathematical morphology
  • Segmentation
  • Watershed

Fingerprint

Dive into the research topics of 'Segmentation and classification of hyperspectral images using watershed transformation'. Together they form a unique fingerprint.

Cite this