ROIseries "Dimensional Juggling"

This example demonstrates:

The dimensional reduction from 4D to 1D
Plotting during this dimensional reduction

Literature:



In [0]:

    
; Set some variables used throughout the script
cloudfree = ['S2A_L2A_UMV32N_20160902T103228_10m_studyarea.tif','S2A_L2A_UMV32N_20161002T103017_10m_studyarea.tif','S2A_L2A_UMV32N_20160922T103357_10m_studyarea.tif','S2A_L2A_UMV32N_20160823T103332_10m_studyarea.tif','S2A_L2A_UMV32N_20160813T103228_10m_studyarea.tif','S2A_L2A_UMV32N_20160624T103023_10m_studyarea.tif','S2A_L2A_UMV32N_20160505T103027_10m_studyarea.tif','S2A_L2A_UMV32N_20160126T104630_10m_studyarea.tif','S2A_L2A_UMV32N_20151207T103733_10m_studyarea.tif','S2A_L2A_UMV32N_20151227T104738_10m_studyarea.tif']
cloudfree = cloudfree[SORT(cloudfree)]
ref = GET_RELDIR('ROIseries_3D__define',1,['data','sentinel_2a'])
shp = ref+"vector\"+"studyarea.shp"
rasterseries = ref + "rasters\" + cloudfree
id_col_name = "Id"
id = "NDVI"
db = FILEPATH(id,/TMP)



In [1]:

    
; Instantiate ROIseries_3D object and add time metadata
NDVI_3D = ROIseries_3D()
NDVI_3D.COOKIE_CUTTER(id,db,shp,id_col_name,rasterseries,SPECTRAL_INDEXER_FORMULA="(R[3] - R[0])/(R[3] + R[0])")
NDVI_3D.TIME_FROM_FILENAMES(rasterseries,[15,4],[19,2],[21,2])









    




images loaded
Pumping up
array indices generated
           1






    




           1

Look at the available keys and make a plot with one of the keys
In the gui: Zoom to see the full data by 1. clicking the magnifier and 2. click and hold somewhere in the visualization and move the mouse around turn on Contours for Z
adjust the 'Z Plane' slider to see contour slices through time



In [2]:

    
print,(NDVI_3D.data).keys()
NDVI_3D.XVOLUME,4,/REVERSE









    




           3
           4
           2
           1






    




% Program caused arithmetic error: Floating illegal operand



In [3]:

    
; A screenshot of the widget:
IMAGE(GET_RELDIR("ROIseries__define",1,["docs","examples","images"])+"XVOLUME.png",/INTERPOLATE)









    












    




IMAGE <6573>

Calculate the average NDVI over time and plot the result.

Compare variation of lightness in the 3D plot with the numeric variantion in 2D:

The time is plottet along the vertical axis in the 3D plot.



In [4]:

    
NDVI_1D_MEAN = NDVI_3D.SPATIAL_MIXER("MEAN")
NDVI_1D_MEAN.unit = LIST("time","Mean of NDVI per object")
NDVI_1D_MEAN.plot(ID = 4)









    












    




Plots will be saved to:C:\Users\keck\AppData\Local\Temp\NDVIplots\
           1



In [5]:

    
; Calculate the GLCM contrast in 45 degrees
NDVI_1D_CON45 = NDVI_3D.SPATIAL_MIXER("GLCM_CON_45")
NDVI_1D_CON45.unit = LIST("time","GLCM_CON_45 of NDVI per object")
NDVI_1D_CON45.plot(ID = 4)









    












    




Plots will be saved to:C:\Users\keck\AppData\Local\Temp\NDVIplots\
           1

Generate 1D features values for the MEAN and CON45 ROIseries_1D objects:

'RAW': Individual values per image-object over time
'MEAN': Average over time per image-object
'STDDEV': Standard Deviation over time per image-object



In [6]:

    
NDVI_1D_MEAN.features_to_csv(['MEAN','STDDEV','RAW'])
NDVI_1D_CON45.features_to_csv(['MEAN','STDDEV','RAW'])









    




Features will be written to: C:\Users\keck\AppData\Local\Temp\NDVI\features\NDVI_features_2017-04-21T15-03-14.77202743291855Z.csv
C:\Users\keck\AppData\Local\Temp\NDVI\features\NDVI_features_2017-04-21T15-03-14.77202743291855Z.csv






    




Features will be written to: C:\Users\keck\AppData\Local\Temp\NDVI\features\NDVI_features_2017-04-21T15-03-14.80304718017578Z.csv
C:\Users\keck\AppData\Local\Temp\NDVI\features\NDVI_features_2017-04-21T15-03-14.80304718017578Z.csv

Conclusion

Have a look at the pathes printed during the last two steps
Theses are the final CSV files holding 0D features to be used in e. g. machine learning