def run(oldpath, newpath):
"""
Runs the 16th basic test suite. Tests:
creating a raster with interpolation using the 'new' non-spatial mode
"""
inputSPD = os.path.join(oldpath, INPUT_SPD)
outputDEM = os.path.join(newpath, OUTPUT_DEM)
data = spatial.readLidarPoints(inputSPD,
classification=lidarprocessor.CLASSIFICATION_GROUND)
(xMin, yMax, ncols, nrows) = spatial.getGridInfoFromData(data['X'], data['Y'],
BINSIZE)
pxlCoords = spatial.getBlockCoordArrays(xMin, yMax, ncols, nrows, BINSIZE)
dem = interpolation.interpGrid(data['X'], data['Y'], data['Z'], pxlCoords,
method='pynn')
iw = spatial.ImageWriter(outputDEM, tlx=xMin, tly=yMax, binSize=BINSIZE)
iw.setLayer(dem)
iw.close()
utils.compareImageFiles(os.path.join(oldpath, OUTPUT_DEM), outputDEM)
def interpGroundReturns(data):
# if given a list of fields, returns a structured array with all of them
ptVals = data.input.getPoints(colNames=['X', 'Y', 'Z', 'CLASSIFICATION'])
# create mask for ground
mask = ptVals['CLASSIFICATION'] == lidarprocessor.CLASSIFICATION_GROUND
# get the coords for this block
pxlCoords = data.info.getBlockCoordArrays()
if ptVals.shape[0] > 0:
# there is data for this block
xVals = ptVals['X'][mask]
yVals = ptVals['Y'][mask]
zVals = ptVals['Z'][mask]
# 'pynn' needs the pynnterp module installed
out = interpolation.interpGrid(xVals,
yVals,
zVals,
pxlCoords,
method='pynn')
# mask out where interpolation failed
invalid = numpy.isnan(out)
out[invalid] = 0
else:
# no data - set to zero
out = numpy.empty(pxlCoords[0].shape, dtype=numpy.float64)
out.fill(0)
out = numpy.expand_dims(out, axis=0)
data.imageOut.setData(out)
def doInterp(minfilteredimage, outinterpimage, controlPtArray):
inImage = gdal.Open(minfilteredimage)
nCols = inImage.RasterXSize
nRows = inImage.RasterYSize
geotransform = inImage.GetGeoTransform()
resolution = geotransform[1]
xmin = numpy.floor(geotransform[0])
xmax = numpy.ceil((xmin + (nCols * resolution)) + resolution)
ymax = numpy.ceil(geotransform[3])
ymin = numpy.floor((ymax - (nRows * resolution)) - resolution)
print(xmin, xmax, ymin, ymax, resolution)
print(geotransform)
proj = inImage.GetProjection()
nulVal = inImage.GetRasterBand(1).GetNoDataValue()
z = ((inImage.GetRasterBand(1)).ReadAsArray())
x = ((inImage.GetRasterBand(2)).ReadAsArray())
y = ((inImage.GetRasterBand(3)).ReadAsArray())
filteredOutMask = (z == nulVal)
mz = numpy.ma.array(z, mask=filteredOutMask).compressed()
mx = numpy.ma.array(x, mask=filteredOutMask).compressed()
my = numpy.ma.array(y, mask=filteredOutMask).compressed()
# Generate a regular grid to interpolate the data.
xi = numpy.linspace(xmin, xmax - resolution, nCols)
yi = numpy.linspace(ymin, ymax - resolution, nRows)
xi, yi = numpy.meshgrid(xi, yi)
# ## Interpolate the values of z for all points in the rectangular grid
# ## Method 1 - Interpolate by matplotlib delaunay triangularizatio and nearest neigh. PLEASE NOTE! THIS FAILS QUITE OFTEN (http://matplotlib.org/api/mlab_api.html#matplotlib.mlab.griddata) But there might be a solution - install mpl_toolkits.natgrid (http://matplotlib.org/mpl_toolkits/)
# #zi = ml.griddata(mx,my,mz,xi,yi,interp='nn') #interpolation is 'nn' by default (natural neighbour based on delaunay triangulation) but 'linear' is faster (see http://matplotlib.1069221.n5.nabble.com/speeding-up-griddata-td20906.html)
#surfaceImg = numpy.flipud(zi)
#use the pylidar natural neighbour interpolation routine
surfaceImg = interpolation.interpGrid(mx, my, mz, (xi, yi), 'pynn')
surfaceImg = numpy.flipud(surfaceImg)
#elif cmdargs.interp == 'mann':
#zi = ml.griddata(mx,my,mz,xi,yi,interp='nn') #interpolation is 'nn' by default (natural neighbour based on delaunay triangulation) but 'linear' is faster (see http://matplotlib.1069221.n5.nabble.com/speeding-up-griddata-td20906.html)
#surfaceImg = numpy.flipud(zi)
saveImage(outinterpimage, surfaceImg, xmin, ymax, resolution, proj, nulVal)
xyAtControl = controlPtArray[:, :2]
zAtControl = controlPtArray[:, 2]
interpzAtControl = pynninterp.NaturalNeighbourPts(mx.astype(numpy.float64),
my.astype(numpy.float64),
mz.astype(numpy.float64),
xyAtControl)
zDiff = zAtControl - interpzAtControl
outf = open('controlCompare_noMed_20151125.csv', 'w')
for i in range(len(zAtControl)):
outf.write("{},{},{}\n".format(interpzAtControl[i], zAtControl[i],
zDiff[i]))
outf.close()