User Tools

Site Tools


hsimaging:hs_iman_howto

Look@MOSI

Look@MOSI is a software for the analysis of hyperspectral images obtained by the MOSI system. It offers functions that are common in commercially or freely available programs for hyper-spectral data analysis. Additionally, Look@MOSI offers features that are not that usual, such as the possibility to pixel-wise calculate 2nd and 4th derivatives of smoothed spectra, or their base-10 logarithm, and display them as an image.

Some details of the program can be found in the paper by Polerecky et al. (2009) published in Applied and Environmental Microbiology, which describes the MOSI system. You can download the published PDF (need subscription), Supplementary material (free).

More results obtained by the program can be seen in papers listed here.

Screenshots

See a gallery of images generated by the Look@MOSI software.

Download

If you are interested in using the program, you can download it from here.

How to use the program

This is only a very quick introduction to the program, sort of like a crash-course.

  1. Load a BIP file.
  2. Select band numbers of the R, G and B channels.
  3. Click on the R, G or B eye button to view the image in the corresponding channel. Double left-click on the image to rescale the colormap, right-click on the image to save it into BMP. Note that smoothed values of the spectral signal can also be plotted.
  4. If you want to define a reference region, click on the corresponding checkbox to enable this definition. Then click on the Open button, and define reference region dimensions. Note: use left=0 and right=480 in version 0.43+, top and bottom can be chosen arbitrarily within the image dimensions. Then click on the Save reference button to calculate vertically averaged spectra in the region and save them into a BIP file. If you have previously saved some other reference regions, click on the Load reference button to load it. Finally, click OK. Leave the checkbox in the main window checked if you want to display spectra normalised to the reference spectra, leave it unchecked if you want to display directly measured spectra.
  5. Click on the Show spectral viewer button (the icon with the graph and red/blue lines) to display the spectrum in a selected pixel. Observe the changing shape of the spectra as you move with the mouse over the R, G or B image (see 3). Double left-click on the Spectral viewer graph to rescale it.
  6. Click on the AVG button to define spectral vectors averaged over a specified area. Specify the boundaries of the area by typing them into the corresponding fields or use the mouse (left-click + drag and release) over the R, G or B image (see 3) to define it interactively. Then, click on the Average spectra button to display the average spectrum. Then click on the Save button to save it into a DAT file. Check the checkbox if you want to display more than one spectrum in the graph.
  7. Explore the HS image as described in points 4-5 to identify pure spectra (or so called endmembers). Save them into separate files.
  8. If you want to fit spectra in each pixel by a set of end-member spectra, click on the VEC button to specify the filenames of the pure spectral vectors. Check the corresponding checkbox to be able to define a pure spectral vector and define its filename through clicking on the Open file button. Check the first checkbox, then define the filename, then check the second checkbox and define the second filename, etc. Do not skip the vectors, i.e., do not define for example v0 and v2 with v1 being undefined! Click OK when you are ready with the definition of all pure spectral vectors to load them. Observe how these spectra are also being displayed while you move the mouse over the R, G or B image, as described in 4. The fitting spectrum, given by the linear combination of the pure spectra (the coefficients c0, c1, … are also displayed), is displayed (gray) along with the measured spectrum (white).
  9. Once the fitted spectra are satisfactory, click on the Decompose HS image button to decompose the measured spectra into pure spectra in each pixel. This will effectively calculate the images of coefficients c0, c1, … (see 7) characterizing the “intensity” distribution of the pure spectra across the image. To view these coefficients, click on the corresponding Show pure spectrum distribution buttons. Double left-click on the image to rescale the colormap, right-click to save it into a BMP file.
  10. The program also allows you to view 2nd and 4th derivatives of the spectra. To do this, select Display derivative to 2 or 4 and click on the “eye button” to display spectral derivative. Note that you may need to change the scale of the image (not the spectra, as this remains the same) to see it correctly. When you want to display the image derived from the original spectra, select 0 derivative.

New features, bug-fixes

Follow this link.


Back to main page.

hsimaging/hs_iman_howto.txt · Last modified: 2013/09/10 14:23 by lpolerec

Page Tools