Classification

Classification Programs and Files



index:

Programs
classification.pdf The classification paper (preliminary)
classify_win.exe windows classification program
source.zip archive of C source code
        Note: requires the GD library.
        See http://www.boutell.com/gd/ for details.


Image and Data Files
small_jbrec5.png band1 image file
small_jbrec10.png band2 image file
small_jbrec15.png band3 image file
small_jbrec21.png band4 image file
small_jbrec30.png band5 image file
small_jbrec40.png band6 image file
small_jbrec50.png band7 image file
small_prior.png prior image file
cov.est.zip Means/Covariance Matrices for each class


Example classification:

    The area to be classified along the Salmon River,ID.



    Enlarged area.



    Results using a uniform prior for yellow starthistle and annual grass. Large outlined areas in the upper portion of the image represent areas where YST is expected. Smaller outlined regions within the same image area represent training sites for YST. Outlined areas at the bottom of the image are grass and river sites.

    YST





    AGRASS




    Prior probability map for yellow starthistle developed from the landscape variables slope, aspect and land use. The darker region in lower left corner represents forest ground while the rest of the image is categoriezed as mixed rangeland.



    The resulting posterior probability map for YST.


    Comparision of classification using the uniform (upper righthand corner) and spatial YST priors. Maps were hardened (given discrete classifications) at p(YST) =.30. Yellow regions are classified as YST, light green as AGRASS, dark green as forest and blue as other.