def createAmp(fi):
(x, y, trans, proj, data) = saa.read_gdal_file(saa.open_gdal_file(fi))
ampdata = np.sqrt(data)
outfile = fi.replace('.tif', '-amp.tif')
print(outfile)
saa.write_gdal_file_float(outfile, trans, proj, ampdata)
return outfile
def fix_geotiff_locations(dir="PRODUCT"):
back = os.getcwd()
os.chdir(dir)
for myfile in glob.glob("*.tif"):
x1, y1, t1, p1, data = saa.read_gdal_file(saa.open_gdal_file(myfile))
easting = t1[0]
resx = t1[1]
rotx = t1[2]
northing = t1[3]
roty = t1[4]
resy = t1[5]
easting = easting + resx / 2.0
northing = northing + resy / 2.0
t1 = [easting, resx, rotx, northing, roty, resy]
tmpfile = "tmp_tiff_{}.tif".format(os.getpid())
if "dem" in myfile or "DEM" in myfile:
saa.write_gdal_file(tmpfile, t1, p1, data)
elif "ls_map" in myfile or "LS" in myfile:
saa.write_gdal_file_byte(tmpfile, t1, p1, data)
else:
saa.write_gdal_file_float(tmpfile, t1, p1, data, nodata=0)
gdal.Translate(myfile,
tmpfile,
metadataOptions=['AREA_OR_POINT=Point'],
noData="0")
os.remove(tmpfile)
os.chdir(back)
def smooth_dem_tiles(demdir, build=True):
os.chdir(demdir)
for mytif in glob.glob("*_8m_dem.tif"):
newtif = mytif.replace(".tif", "_30m.tif")
print("creating file {}".format(newtif))
gdal.Translate(newtif, mytif, xRes=30, yRes=30)
for mytif in glob.glob("*_8m_dem_30m.tif"):
new_name = mytif.replace(".tif", "_smooth.tif")
print("new_name {}".format(new_name))
if not os.path.isfile(new_name):
print("Cleaning up DEM {}".format(mytif))
src_ds = gdal.Open(mytif)
(x1, y1, trans, proj, data) = saa.read_gdal_file(src_ds)
if src_ds is None:
print('Unable to open %s' % mytif)
sys.exit(1)
srcband = src_ds.GetRasterBand(1)
no_data = srcband.GetNoDataValue()
print("noData value is {}".format(no_data))
dem = mytif.replace(".tif", ".dem")
par = dem + ".par"
ps2dem(mytif, dem, par)
tmp_name = mytif.replace(".tif", "_tmp.dem")
cmd = "fill_gaps {in1} {width} {out} - - 1 100".format(
in1=dem, width=x1, out=tmp_name)
os.system(cmd)
cmd = "data2geotiff {par} {in1} 2 {out}".format(par=par,
in1=tmp_name,
out=new_name)
os.system(cmd)
print("removing {} {} {}".format(dem, par, tmp_name))
os.remove(dem)
os.remove(par)
os.remove(tmp_name)
if build:
cmd = "gdalbuildvrt full_area.vrt *_smooth.tif"
os.system(cmd)
cmd = "gdal_translate full_area.vrt full_area.tif"
os.system(cmd)
cmd = "ps2dem.py full_area.tif full_area.dem full_area.dem.par"
os.system(cmd)
logging.info("Finished creating output")
return "full_area.dem", "full_area.dem.par"
def get2sigmacutoffs(fi):
(x, y, trans, proj, data) = saa.read_gdal_file(saa.open_gdal_file(fi))
top = np.percentile(data, 98)
data[data > top] = top
stddev = np.std(data)
mean = np.mean(data)
lo = mean - 2 * stddev
hi = mean + 2 * stddev
return lo, hi
def byteSigmaScale(infile, outfile):
lo, hi = get2sigmacutoffs(infile)
print("2-sigma cutoffs are {} {}".format(lo, hi))
gdal.Translate(outfile,
infile,
outputType=gdal.GDT_Byte,
scaleParams=[[lo, hi, 1, 255]],
resampleAlg="average",
noData="0")
# For some reason, I'm still getting zeros in my byte images eventhough I'm using 1,255 scaling!
# The following in an attempt to fix that!
(x, y, trans, proj, data) = saa.read_gdal_file(saa.open_gdal_file(infile))
mask = (data > 0).astype(bool)
(x, y, trans, proj, data) = saa.read_gdal_file(saa.open_gdal_file(outfile))
mask2 = (data > 0).astype(bool)
mask3 = mask ^ mask2
data[mask3 == True] = 1
saa.write_gdal_file_byte(outfile, trans, proj, data, nodata=0)
def get2sigmacutoffs(fi):
(x, y, trans, proj, data) = saa.read_gdal_file(saa.open_gdal_file(fi))
data = data.astype(float)
data[data == 0] = np.nan
top = np.nanpercentile(data, 99)
data[data > top] = top
stddev = np.nanstd(data)
mean = np.nanmean(data)
lo = mean - 2 * stddev
hi = mean + 2 * stddev
return lo, hi
def clean_dem(in_dem, out_dem):
(x, y, trans, proj, data) = saa.read_gdal_file(saa.open_gdal_file(in_dem))
logging.info("Replacing values less than -1000 with zero")
data[data <= -1000] = -32767
logging.info(f"DEM Maximum value: {np.max(data)}")
logging.info(f"DEM minimum value: {np.min(data)}")
if data.dtype == np.float32:
saa.write_gdal_file_float(out_dem, trans, proj, data.astype(np.float32))
elif data.dtype == np.uint16:
saa.write_gdal_file(out_dem, trans, proj, data)
else:
logging.error(f"ERROR: Unknown DEM data type {data.dtype}")
sys.exit(1)
def utm2dem(inDem, outDem, demPar, dataType="float"):
demParIn = "dem_par.in"
dataType = dataType.lower()
basename = os.path.basename(inDem)
logname = basename + "_utm_dem.log"
log = open(logname, "w")
print("UTM DEM in GEOTIFF format: {}".format(inDem))
print("output DEM: {}".format(outDem))
print("output DEM parameter file: {}".format(demPar))
print("log file: {}".format(logname))
(x, y, trans, proj, data) = saa.read_gdal_file(saa.open_gdal_file(inDem))
xsize = x
ysize = y
east = trans[0]
north = trans[3]
pix_east = trans[1]
pix_north = trans[5]
ds = gdal.Open(inDem)
prj = ds.GetProjection()
s = prj.split("[")
for t in s:
if "false_northing" in t:
u = t.split('"')
v = u[2].split(",")
w = v[1].split("]")
false_north = w[0]
print("found false_north {}".format(false_north))
srs = osr.SpatialReference(wkt=prj)
string = srs.GetAttrValue('projcs')
t = string.split(" ")
zone = t[5]
print("Found zone string {} of length {}".format(zone, len(zone)))
if len(zone) == 3:
zone = zone[0:2]
else:
zone = zone[0]
print("found zone {}".format(zone))
src = gdal.Open(inDem, gdalconst.GA_ReadOnly)
string = src.GetMetadata()
pixasarea = string["AREA_OR_POINT"]
if "AREA" in pixasarea:
print("Pixel as Area! Updating corner coordinates to pixel as point")
print("pixel upper northing (m): {} easting (m): {}".format(
north, east))
east = east + pix_east / 2.0
north = north + pix_north / 2.0
print("Update pixel upper northing (m): {} easting (m): {}".format(
north, east))
pix_size = pix_east
print("approximate DEM latitude pixel spacing (m): {}".format(pix_size))
# Create the input file for create_dem_par
f = open(demParIn, "w")
f.write("UTM\n")
f.write("WGS84\n")
f.write("1\n")
f.write("{}\n".format(zone))
f.write("{}\n".format(false_north))
f.write("{}\n".format(basename))
if "float" in dataType:
f.write("REAL*4\n")
elif "int16" in dataType:
f.write("INTEGER*2\n")
f.write("0.0\n")
f.write("1.0\n")
f.write("{}\n".format(xsize))
f.write("{}\n".format(ysize))
f.write("{} {}\n".format(pix_north, pix_east))
f.write("{} {}\n".format(north, east))
f.close()
# Create a new dem par file
if os.path.isfile(demPar):
os.remove(demPar)
execute("create_dem_par {} < {}".format(demPar, demParIn), logfile=log)
# Replace 0 with 1; Replace anything <= -32767 with 0; byteswap
data[data == 0] = 1
data[data <= -32767] = 0
data = data.byteswap()
# Convert to ENVI (binary) format
tmptif = "temporary_dem_file.tif"
if "float" in dataType:
saa.write_gdal_file_float(tmptif, trans, proj, data.astype(np.float32))
elif "int16" in dataType:
saa.write_gdal_file(tmptif, trans, proj, data)
gdal.Translate(outDem, tmptif, format="ENVI")
os.remove(tmptif)
os.remove(outDem + ".aux.xml")
filename, file_extension = os.path.splitext(outDem)
os.remove(outDem.replace(file_extension, ".hdr"))
def report_min(in_dem):
(x, y, trans, proj, data) = saa.read_gdal_file(saa.open_gdal_file(in_dem))
logging.debug(f"DEM file {in_dem} minimum is {np.min(data)}")
def get_dem(x_min, y_min, x_max, y_max, outfile, post=None, processes=1, dem_name=None, leave=False, dem_type='utm'):
if post is not None:
logging.info(f"Snapping to grid at posting of {post} meters")
if y_min < -90 or y_max > 90:
raise ValueError(f"Please use latitude in range (-90, 90) ({y_min}, {y_max})")
if x_min > x_max:
logging.warning("WARNING: minimum easting > maximum easting - swapping")
(x_min, x_max) = (x_max, x_min)
if y_min > y_max:
logging.warning("WARNING: minimum northing > maximum northing - swapping")
(y_min, y_max) = (y_max, y_min)
# Figure out which DEM and get the tile list
(demname, demproj, tile_list, poly_list) = get_best_dem(y_min, y_max, x_min, x_max, dem_name=dem_name)
demproj = int(demproj)
logging.info(f"demproj is {demproj}")
# Add buffer for REMA
if 'REMA' in demname or 'GIMP' in demname:
x_min -= 4
x_max += 4
if 'EU_DEM' in demname:
y_min -= 2
y_max += 2
# Copy the files into a dem directory
if not os.path.isdir("DEM"):
os.mkdir("DEM")
# Download tiles in parallel
logging.info("Fetching DEM tiles to local storage")
p = mp.Pool(processes=processes)
p.map(
get_tile_for,
[(demname, fi) for fi in tile_list]
)
p.close()
p.join()
# os.system("gdalbuildvrt temp.vrt DEM/*.tif")
if "SRTMGL" in demname:
nodata = -32768
elif "GIMP" in demname:
nodata = None
elif "REMA" in demname:
nodata = 0
elif "NED" in demname or "EU_DEM_V11" in demname:
nodata = -3.4028234663852886e+38
else:
raise DemError(f'Unable to determine NoData value for DEM {demname}')
write_vrt(demproj, nodata, tile_list, poly_list, 'temp.vrt')
#
# Set the output projection to either NPS, SPS, or UTM
#
if demproj == 3413: # North Polar Stereo
outproj = 'EPSG:3413'
outproj_num = 3413
elif demproj == 3031: # South Polar Stereo
outproj = 'EPSG:3031'
outproj_num = 3031
else:
lon = (x_max + x_min) / 2
zone = math.floor((lon + 180) / 6 + 1)
if zone > 60:
zone -= 60
if (y_min + y_max) / 2 > 0:
outproj = ('EPSG:326%02d' % int(zone))
outproj_num = int("326%02d" % int(zone))
else:
outproj = ('EPSG:327%02d' % int(zone))
outproj_num = int("327%02d" % int(zone))
tmpdem = "xxyyzz_img.tif"
tmpdem2 = "aabbcc_img.tif"
tmpproj = "lmnopqr_img.tif"
if os.path.isfile(tmpdem):
logging.info(f"Removing old file {tmpdem}")
os.remove(tmpdem)
if os.path.isfile(tmpproj):
logging.info("Removing old file projected dem file")
os.remove(tmpproj)
pixsize = 30.0
gcssize = 0.00027777777778
if demname == "SRTMGL3":
pixsize = 90.
gcssize *= 3
if demname == "NED2":
pixsize = 60.
gcssize *= 2
logging.info("Creating initial raster file")
logging.info(f" tmpdem {tmpdem}")
logging.info(f" pixsize {pixsize}")
logging.info(f" bounds: x_min {x_min}; y_min {y_min}; x_max {x_max}; y_max {y_max}")
# xform bounds to projection of the DEM
if demproj != 4326:
transformer = Transformer.from_crs('epsg:4326', f'epsg:{demproj}')
t_x, t_y = transformer.transform([x_min, x_max], [y_min, y_max])
x_min, x_max = sorted(t_x)
y_min, y_max = sorted(t_y)
logging.info(f" transformed bounds: x_min {x_min}; y_min {y_min}; x_max {x_max}; y_max {y_max}")
if demproj == 4269 or demproj == 4326:
res = gcssize
else:
res = pixsize
gdal.Warp(tmpdem, "temp.vrt", xRes=res, yRes=res, outputBounds=[x_min, y_min, x_max, y_max],
resampleAlg="cubic", dstNodata=-32767)
# If DEM is from NED collection, then it will have a NAD83 ellipse -
# need to convert to WGS84
# Also, need to convert from pixel as area to pixel as point
if "NED" in demname:
logging.info("Converting to WGS84")
gdal.Warp("temp_dem_wgs84.tif", tmpdem, dstSRS="EPSG:4326")
logging.info("Converting to pixel as point")
x1, y1, t1, p1, data = \
saa.read_gdal_file(saa.open_gdal_file("temp_dem_wgs84.tif"))
lon = t1[0]
resx = t1[1]
rotx = t1[2]
lat = t1[3]
roty = t1[4]
resy = t1[5]
lon = lon + resx / 2.0
lat = lat + resy / 2.0
t1 = [lon, resx, rotx, lat, roty, resy]
saa.write_gdal_file_float(tmpdem, t1, p1, data)
if not leave:
os.remove("temp_dem_wgs84.tif")
clean_dem(tmpdem, tmpdem2)
shutil.move(tmpdem2, tmpdem)
gdal.Translate(tmpdem2, tmpdem, metadataOptions=['AREA_OR_POINT=Point'])
shutil.move(tmpdem2, tmpdem)
# Reproject the DEM file into UTM space
if demproj != outproj_num:
logging.info(f"Translating raster file to projected coordinates ({outproj})")
gdal.Warp(tmpproj, tmpdem, dstSRS=outproj, xRes=pixsize, yRes=pixsize, resampleAlg="cubic",
srcNodata=-32767, dstNodata=-32767)
infile = tmpproj
else:
infile = tmpdem
report_min(infile)
# Snap to posting grid
if post:
snap_to_grid(post, pixsize, infile, outfile)
else:
shutil.copy(infile, outfile)
report_min(outfile)
# Clean up intermediate files
if not leave:
if os.path.isfile(tmpdem):
logging.info(f"Removing temp file {tmpdem}")
os.remove(tmpdem)
if os.path.isfile(tmpproj):
logging.info(f"Removing temp file {tmpproj}")
os.remove(tmpproj)
logging.info("Successful Completion!")
if dem_type.lower() == 'utm':
return demname
elif dem_type.lower() == 'latlon':
pixsize = 0.000277777777778
gdal.Warp(
"temp_dem.tif", outfile, dstSRS="EPSG:4326", xRes=pixsize, yRes=pixsize, resampleAlg="cubic",
dstNodata=-32767
)
shutil.move("temp_dem.tif", outfile)
elif dem_type.lower() == 'isce':
pixsize = 0.000277777777778
gdal.Warp("temp_dem.tif", outfile, format="ENVI", dstSRS="EPSG:4326", xRes=pixsize, yRes=pixsize,
resampleAlg="cubic", dstNodata=-32767)
shutil.move("temp_dem.tif", outfile)
hdr_name = os.path.splitext(outfile)[0] + ".hdr"
dem2isce.dem2isce(outfile, hdr_name, f'{outfile}.xml')
else:
raise NotImplementedError(f'Cannot get DEM for unkown type {dem_type}')
return demname
def makeColorPhase(inFile,
rateReduction=1,
shift=0,
ampFile=None,
scale=0,
table='CMY'):
samples = 1024
pinf = float('+inf')
ninf = float('-inf')
# fnan = float('nan')
mod2pi = False
if table == 'CMY':
mod2pi = True
R, G, B = makeCycleColor(samples)
elif table == 'RYB':
R, G, B = makeContinuousColor(samples)
elif table == 'RWB':
R, G, B = makeRWBColor(samples)
else:
print("ERROR: Unknown color table: {}".format(table))
exit(1)
#
# Read in the phase data
#
x, y, trans, proj = saa.read_gdal_file_geo(saa.open_gdal_file(inFile))
# If data if too big, resize it
if x > 4096 or y > 4096:
phaseTmp = "{}_small.tif".format(
os.path.basename(inFile.replace(".tif", "")))
gdal.Translate(phaseTmp, inFile, height=4096)
x, y, trans, proj, data = saa.read_gdal_file(
saa.open_gdal_file(phaseTmp))
print("Created small tif of size {} x {}".format(x, y))
else:
x, y, trans, proj, data = saa.read_gdal_file(
saa.open_gdal_file(inFile))
print("Using full size tif of size {} x {}".format(x, y))
phaseTmp = inFile
# Make a black mask for use after colorization
mask = np.ones(data.shape, dtype=np.uint8)
mask[data[:] == 0] = 0
# Scale to 0 .. samples-1
data[:] = data[:] + shift
if mod2pi == True:
data[:] = data[:] % (2 * rateReduction * np.pi)
const = samples / (2 * rateReduction * np.pi)
data[:] = data[:] * const
else:
mask = np.ones(data.shape, dtype=np.uint8)
mask[data == pinf] = 0
mask[data == ninf] = 0
mask[np.isnan(data)] = 0
data[mask == 0] = 0
# mini = np.min(data)
# maxi = np.max(data)
mini = np.percentile(data, 2)
maxi = np.percentile(data, 98)
data[data < mini] = mini
data[data > maxi] = maxi
data[:] = (data[:] - mini) / (maxi - mini)
data[:] = data * float(samples)
print(np.max(data))
print(np.min(data))
hist = np.histogram(data)
print(hist[1])
print(hist[0])
data[data == samples] = samples - 1
# Convert to integer for indexing
idata = np.zeros(data.shape, dtype=np.uint16)
idata[:] = data[:]
# Make the red, green, and blue versions
red = np.zeros(data.shape, dtype=np.uint8)
green = np.zeros(data.shape, dtype=np.uint8)
blue = np.zeros(data.shape, dtype=np.uint8)
red = R[idata[:]]
green = G[idata[:]]
blue = B[idata[:]]
# Apply the black mask
red[mask == 0] = 0
green[mask == 0] = 0
blue[mask == 0] = 0
if ampFile is None:
# Write out the RGB phase image
fileName = inFile.replace(".tif", "_rgb.tif")
saa.write_gdal_file_rgb(fileName, trans, proj, red, green, blue)
# If we have amplitude, use that
else:
# Make the red, green, and blue floating point versions
redf = np.zeros(data.shape)
greenf = np.zeros(data.shape)
bluef = np.zeros(data.shape)
# Scale from 0 .. 1
redf[::] = red[::] / 255.0
greenf[::] = green[::] / 255.0
bluef[::] = blue[::] / 255.0
# Read in the amplitude data
x1, y1, trans1, proj1 = saa.read_gdal_file_geo(
saa.open_gdal_file(ampFile))
# If too large, resize the data
if x1 > 4096 or y1 > 4096:
ampTmp = "{}_small.tif".format(
os.path.basename(ampFile.replace(".tif", "")))
gdal.Translate(ampTmp, ampFile, height=y, width=x)
x1, y1, trans1, proj1, amp = saa.read_gdal_file(
saa.open_gdal_file(ampTmp))
else:
x1, y1, trans1, proj1, amp = saa.read_gdal_file(
saa.open_gdal_file(ampFile))
ampTmp = ampFile
if (x != x1) or (y != y1):
cutFiles([phaseTmp, ampTmp])
# if phaseTmp != inFile:
# os.remove(phaseTmp)
phaseTmp = phaseTmp.replace(".tif", "_clip.tif")
x, y, trans, proj, data = saa.read_gdal_file(
saa.open_gdal_file(phaseTmp))
# if ampTmp != ampFile:
# os.remove(ampTmp)
ampTmp = ampTmp.replace(".tif", "_clip.tif")
x1, y1, trans1, proj1, amp = saa.read_gdal_file(
saa.open_gdal_file(ampTmp))
print("Data shape is {}".format(data.shape))
print("Amp shape is {}".format(amp.shape))
# Make a black mask for use after colorization
mask = np.ones(amp.shape, dtype=np.uint8)
mask[amp == pinf] = 0
mask[amp == ninf] = 0
mask[np.isnan(amp)] = 0
amp[mask == 0] = 0
ave = np.mean(amp)
print("Mean of amp data is {}".format(ave))
amp[mask == 0] = ave
print("AMP HISTOGRAM:")
hist = np.histogram(amp)
print(hist[1])
print(hist[0])
ave = np.mean(amp)
print("Amp average is {}".format(ave))
print("Amp median is {}".format(np.median(amp)))
print("Amp stddev is {}".format(np.std(amp)))
# Rescale amplitude to 2-sigma byte range, otherwise may be all dark
amp2File = createAmp(ampTmp)
myrange = get2sigmacutoffs(amp2File)
newFile = "tmp.tif"
gdal.Translate(newFile,
amp2File,
outputType=gdal.GDT_Byte,
scaleParams=[myrange],
resampleAlg="average")
x, y, trans, proj, amp = saa.read_gdal_file(
saa.open_gdal_file(newFile))
# if ampTmp != ampFile:
# os.remove(ampTmp)
# os.remove(amp2File)
# os.remove(newFile)
print("2-sigma AMP HISTOGRAM:")
hist = np.histogram(amp)
print(hist[1])
print(hist[0])
# Scale amplitude from 0.0 to 1.0
ampf = np.zeros(data.shape)
ampf = amp / 255.0
ampf = ampf + float(scale)
ampf[ampf > 1.0] = 1.0
print("SCALED AMP HISTOGRAM:")
hist = np.histogram(ampf)
print(hist[1])
print(hist[0])
# Perform color transformation
h = np.zeros(data.shape)
l = np.zeros(data.shape)
s = np.zeros(data.shape)
for j in range(x):
for i in range(y):
h[i,
j], l[i,
j], s[i,
j] = colorsys.rgb_to_hls(redf[i, j],
greenf[i, j], bluef[i,
j])
print("LIGHTNESS HISTOGRAM:")
hist = np.histogram(l)
print(hist[1])
print(hist[0])
l = l * ampf
print("NEW LIGHTNESS HISTOGRAM:")
hist = np.histogram(l)
print(hist[1])
print(hist[0])
for j in range(x):
for i in range(y):
redf[i, j], greenf[i, j], bluef[i, j] = colorsys.hls_to_rgb(
h[i, j], l[i, j], s[i, j])
red = redf * 255
green = greenf * 255
blue = bluef * 255
print("TRANFORMED RED HISTOGRAM:")
hist = np.histogram(red)
print(hist[1])
print(hist[0])
# Apply mask
red[mask == 0] = 0
green[mask == 0] = 0
blue[mask == 0] = 0
# Write out the RGB phase image
fileName = inFile.replace(".tif", "_amp_rgb.tif")
saa.write_gdal_file_rgb(fileName, trans, proj, red, green, blue)
#
# This code makes an image to show off the color table
#
# rainbow_red = np.zeros((1024,1024),np.uint8)
# rainbow_green = np.zeros((1024,1024),np.uint8)
# rainbow_blue = np.zeros((1024,1024),np.uint8)
# for i in range(1024):
# for j in range(1024):
# idx = int((float(i)/1024.0)*samples)
# rainbow_red[i,j] = R[idx]
# rainbow_green[i,j] = G[idx]
# rainbow_blue[i,j] = B[idx]
# saa.write_gdal_file_rgb("rainbow.tif",trans,proj,rainbow_red,rainbow_green,rainbow_blue)
return (fileName)
def makeChangeBrowse(geotiff, type="MSCD"):
# read in the data
x, y, trans, proj, data = saa.read_gdal_file(saa.open_gdal_file(geotiff))
red = np.zeros(data.shape, dtype=np.uint8)
blue = np.zeros(data.shape, dtype=np.uint8)
green = np.zeros(data.shape, dtype=np.uint8)
newData = np.zeros(data.shape, dtype=np.uint8)
if type == "SACD":
#
# Make the greyscale image
#
lut = [0, 64, 0, 192]
#
# Make the color images
#
red_lut = [0, 255, 0, 1]
green_lut = [0, 1, 0, 1]
blue_lut = [0, 1, 0, 255]
for i in range(y):
for j in range(x):
newData[i, j] = lut[data[i, j]]
red[i, j] = red_lut[data[i, j]]
green[i, j] = green_lut[data[i, j]]
blue[i, j] = blue_lut[data[i, j]]
else:
#
# get data median and histogram
#
median = np.median(data)
bins = np.zeros(MAX_CLASSES, dtype=np.int8)
for i in range(MAX_CLASSES):
bins[i] = i
hist = np.histogram(data, bins=bins)
#
# count the number of classes present in histogram and set class number,
# making the median class 0, keeping all zeors as class 0, and all others
# with histogram values to a linear sequence 1, 2, 3, ...
#
class_cnt = 0
next_class = 1
classifications = np.zeros(MAX_CLASSES, dtype=np.int8)
classifications[:] = -1
classifications[0] = 0
for i in range(0, len(hist[0])):
if hist[0][i] != 0:
class_cnt = class_cnt + 1
if hist[1][i] == 0:
# we have zeros in the image - need to be handled as background
classifications[i] = 0
elif i == median:
classifications[i] = 0
else:
classifications[i] = next_class
next_class = next_class + 1
#
# Make LUT to map classifications to greyscale values
# Start at 64, increment by 192/(#classes-2) to get
# sequences like {64,255}, {64,160,255}, {64,128,192,256}, etc,
# always leaving the median class and zero pixels as zero valued
#
lut = np.zeros(class_cnt, dtype=np.uint8)
if class_cnt == 1:
print("ERROR: Only found one class")
exit(1)
if (class_cnt == 2):
lut[0] = 0
lut[1] = 255
else:
val = 64
inc = 192 / (class_cnt - 2)
for i in range(class_cnt):
if i != median and hist[1][i] != 0:
lut[classifications[i]] = int(val)
val = val + inc
if val > 255:
val = int(255)
#
# Use the look up table to set the values in newData array
#
newData = lut[classifications[data]]
#
# Create the color version of the data
# Here, we use the same classifications as
# an index into a color look up table.
#
red_lut = [1, 255, 1, 1, 255, 255, 1, 128, 128, 1]
green_lut = [1, 1, 1, 255, 128, 1, 255, 255, 1, 128]
blue_lut = [1, 1, 255, 1, 1, 128, 128, 1, 255, 255]
for i in range(y):
for j in range(x):
k = classifications[data[i, j]]
red[i, j] = red_lut[k]
blue[i, j] = blue_lut[k]
green[i, j] = green_lut[k]
#
# Write out the greyscale png files
#
outName = geotiff.replace(".tif", "_byte.tif")
pngName = geotiff.replace(".tif", "_byte_full.png")
saa.write_gdal_file_byte(outName, trans, proj, newData.astype(np.byte))
gdal.Translate(pngName,
outName,
format="PNG",
outputType=gdal.GDT_Byte,
scaleParams=[[0, 255]],
noData="0 0 0")
os.remove(outName)
#
# Write out the RGB tif
#
outName = geotiff.replace(".tif", "_rgb.tif")
pngName = geotiff.replace(".tif", "_rgb_full.png")
saa.write_gdal_file_rgb(outName, trans, proj, red, green, blue)
gdal.Translate(pngName,
outName,
format="PNG",
outputType=gdal.GDT_Byte,
scaleParams=[[0, 255]],
noData="0 0 0")
#
# Make the ASF standard browse and kmz images
#
tmpName = geotiff.replace(".tif", "_rgb")
makeAsfBrowse(outName, tmpName, use_nn=True)
os.remove(outName)