Você está na página 1de 3

MultiSpec A MULTISPECTRAL IMAGE DATA ANALYSIS SYSTEM

MultiSpec (PurdueResearch Foundation) is aprocessingsystem forinteractively analyzing Earth


observationalmultispectralimagedatasuchasthatproducedbytheLandsatseriesofEarthsatellitesand
hyperspectralimagedatafromcurrentandfutureairborneandspacebornesystemssuchasAVIRIS.The
primaryobjectiveofMultiSpecisasanaidtoexporttheresultsofourresearchintodevisinggood
methods for analyzing such hyperspectral image data. It has also found significant use in other
applicationssuchasmultibandmedicalimageryandinK12anduniversityleveleducationalactivities.
Therearecurrentlyinexcessofseveralthousandknown,registeredusers.
MultiSpecsatisfiesthefollowingdesigngoals:
Theimplementationshouldbeonareadilyavailableplatformwhichhasadequateprocessing
power,butwhichisfinanciallywithinthereachofanyremotesensinguser(i.e.,$2000).
Thesystemshouldbeeasytolearnandeasytouse,evenfortheinfrequentuser,usingthemost
modernofsoftwareenvironments.
Thesystemshouldprovideforeasyimportofdatainavarietyofformats,andeasyexportof
results,bothinthematicmapandintabularform.
TheworkofbuildingthecurrentcapabilitybeganbyimplementinganupgradedversionoftheLARSYS
multispectralimagedataanalysissystem.LARSYSisoneofthefirstremotesensingmultispectraldata
processing systems, originally created during the 1960's. A number of systems in government
laboratories, universityresearchlabs,andseveralcommercially offeredproductsaredescendants of
LARSYS.Thecurrentsystem,calledMultiSpec,hasbeenimplementedfortheAppleMacintoshand
PCWindowspersonalworkstations.Areasonablycurrentgeneration,middlerangemachineandcolor
display,wouldhaveastreetpriceoflessthan$2000atthepresenttime.Suchasystemiscapableof
classifyinginexcessof6millionpixelclassesperminuteusing12bandsandaGaussianmaximum
likelihood scheme.Giventhecurrentcost/performancetrends,evenmorecosteffective systems are
likelytobeavailableinthefuture.
NewcapabilitiesarecontinuallyaddedtoMultiSpecastheyemergefromourresearchonhyperspectral
processing.CapabilitiesofthelatestversionofMultiSpecincludethefollowing.

Importdata ineitherBinaryorASCIIformatwithorwithoutaheader,andinBand
InterleavedbyLine(BIL),BandSequential(BSQ),orBandInterleavedbySample(BIS)
formats.Thedatavaluesmaybe8bitinteger,16bitinteger,32bitinteger,32bitrealor
64bitreal.Incasesoftwo,fouroreightbytespersample,thebytesmaybeineitherorder.

DisplaymultispectralimagesinavarietyofB/Worcolorformatsusinglinearorequal
areagrayscales;display(internallygenerated)thematicimagesalsoinB/Worcolor,with
anabilitytocontrolthecolorusedforeachtheme.

Histogramdataforuseindeterminingthegrayscaleregimeforadisplayorforlistingand
graphing.

MultiSpec Purdue Research Foundation, Inc.


For further information, contact
Larry Biehl
Purdue University ITaP/RCAC
West Lafayette, IN 47907-1285
Internet: biehl@purdue.edu

- 1 -

11/8/2016

Reformatthedatafileinanumberofways,e.g.,byaddingastandardheader,changing
fromanyoneofthethreeinterleaveformatstoeitheroftheothertwo,editingoutchannels,
combiningfiles,addingormodifyingchanneldescriptions,mosaicingdatasets,changing
thegeometryofadataset,andanumberofotherchanges.

Createnewchannelsofdatafromexistingchannels.Thenewchannelsmaybetheresult
ofaprincipalcomponentsorfeatureextractiontransformationoftheexistingones,orthey
mayresultfromtheratioofalinearcombinationofexistingbandsdividedbyadifferent
linearcombinationofbands.

Clusterdatausingeitherasinglepassoraniterative(isodata)clusteringalgorithm.Save
the results fordisplay as a thematic map. Cluster statistics can also besaved as class
statistics. Initial cluster centers may be prespecified or established by one of several
algorithms.UseofclusteringfollowedbyECHOspectral/spatialclassificationprovidesan
effectivemultivariatescenesegmentationscheme.

Defineclassesviadesignatingrectangular,polygonaltrainingfieldsormaskimagefiles,
computefieldandclassstatistics,anddefinetestfieldsforuseinevaluatingclassification
resultsquantitatively.AfeaturecalledEnhanceStatisticsalsoallowsonetoimprovethe
extenttowhichthedefinedclassstatisticsfitthecompositeofalldatainthedataset.A
covarianceestimationscheme(LOOC)canoptimizethatestimateforsmalltrainingsets.

Determinethebestspectralfeaturestouseforagivenclassificationusing(a)searching
forthebestsubsetoffeaturesusinganyoffivestatisticaldistancemeasures,(b)amethod
based directly upon decision boundaries defined by training samples, or (c) a second
method based directly upon the discriminant functions. Also included are methods
especially designed to search for narrow spectral features such as spectroscopic
characteristics, and for use of projection pursuit as a means of further improving the
featuresextracted.

Classify a designated area in the data file. Six different classification algorithms are
available:useofminimumdistancetomeans,correlation(SAM),matchedfilter(CEM),
Fisherlineardiscriminant,theGaussianmaximumlikelihoodpixelscheme,ortheECHO
spectral/spatialclassifier.Savetheresultsfordisplayasathematicmap,withorwithout
trainingandtestfieldsbeingshown.Applyathresholdtoaclassification,andgeneratea
probabilityorthresholdmapshowingthedegreeofmembershipofeachpixeltotheclass
towhichitwasassigned.

Listclassificationresultsoftrainingortestareasintabularformonaperfield,perclass,
orgroupsofclassesbasis.

Showa graph ofthe spectralvalues ofacurrentlyselected pixelorthemean one


standarddeviationforaselectedarea.Showscatterdiagramsofdatafrompairsofbands
andellipses ofconcentration fortraining setsandselected areas.Show agraphofthe

MultiSpec description

11/8/16

histogramsoftheclassorfielddatavaluesusedfortraining.Showthecoordinatesofa
currentlyselectedarea.

Showacolorpresentationofthecorrelationmatrixforafieldorclassasavisualization
toolespeciallyforhyperspectraldata.

Severaladditional utilityfunctionsincludinglistingoutasubsetofthedatae.g.,foruse
externally,conductingprincipalcomponentanalysis,etc.

Transfer intermediateorfinal results,betheytext,B/Wimageorcolorimage,toother


application programs such as word processors, spreadsheet, or graphics program by
copying and pasting or by saving and then opening the saved file within another
application.

TheMultiSpecimplementationiscarriedoutinsuchawaythattheprimarylimitonthenumberoflines
orcolumnsofthedata,thenumberofspectralbands,etc.,arethosedeterminedbytheavailablediskand
memory space. Taken together, these capabilities provide a stateoftheart capability to analyze
moderateandhighdimensionalmultispectraldatasetsofpracticalsize.Allversions,alongwitha170+
page document listing its capabilities in more detail and providing tutorial exercises in its use, is
available,alongwithsubstantialadditionaldocumentation,viatheWorldWideWebatthefollowing
URL:
https://engineering.purdue.edu/~biehl/MultiSpec/

MultiSpec description

11/8/16

Você também pode gostar