HyperMix

HyperMix is a new toolbox comprising different parts of the hyperspectral unmixing chain, which has been developed using the Orfeo Toolbox (OTB) distributed by CNES and the Fast Light Toolkit (FLTK). In the following, we describe the parts of the spectral unmixing chain that are available in HyperMix:

Endmember estimation:

  • Virtual Dimensionality (VD)
  • Hysime

Dimensional reduction:

  • Principal component analysis (PCA)

Endmember extraction:

  • Algorithms using spectral information: iterative error analysis (IEA), N-FINDR, orthogonal subspace projection (OSP), vertex component analysis (VCA)
  • Algorithms using both spatial and spectral information: automated morpho-logical endmember extraction (AMEE), spatial-spectral endmember extraction (SSEE), spatial pre-processing (SPP)

Abundance estimation:

  • Unconstrained (UCLS), sum-to-one least-squares (SCLS), nonnegativity constraint least-squares (NCLS) and fully constrained least-squares (FCLS)

Figure 1 shows screen captures of the HyperMix OTB open source tool for spectral unmixing.


Figure 1 HyperMix tool.

The tool also includes a database of synthetic hyperspectral images (simulated using fractals). Each fractal image is formed from a set of classes or clusters, and mixed pixels are generated inside each cluster using library signatures obtained from a library of mineral spectral signatures available online from U.S. Geological Survey (USGS) – http://speclab.cr.usgs.govclassification/extraction results by calling top-down operators.


Figure 2 Different views of the HyperMix tool.

System control

As shown in Fig. 1, the HyperMix tool is designed to be easy to use. The tool includes a graphical interface with a main menu in which the algorithm to execute can be chosen. Also the user interfaces includes several displays. ne display show the image data, in the second display we can see the abundances map calculated after the unmixing and the third display shows the endmembers. Fig. 2 provides a screenshoot of the HyperMix tool.

Publications: 

L. I. Jimenez, G. Martin and A. Plaza. HyperMix: A New Tool for Quantitative Evaluation of Endmember Identification and Spectral Unmixing Techniques. IEEE Geoscience and Remote Sensing Symposium (IGARSS'12), Munich, Germany, 2012.

G. Martin, L. I. Jimenez and A. Plaza. A New Tool for Evaluating Spectral Unmixing Applications for Remotely Sensed Hyperspectral Image Analysis. International Conference on Geographic Object-Based Image Analysis (GEOBIA), Rio de Janeiro, Brazil, 2012.