def classification(data):
# Add channel
pcimod(file=inputfile, pciop='add', pcival=[0, 0, 1, 0, 0, 0])
# Define input parameters
classesNumber = [8] # define number of classes
iterations = [10] # define number iterations
moveThresh = [0.01] # define move threshhold
print "Running unsupervized k-means classification..."
print "Creating " + str(classesNumber) + " classes, applying " + str(
iterations) + " iterations at a move-threshhold of " + str(
moveThresh)
# Run algorithm and create classification report
try:
Report.clear()
enableDefaultReport(report)
kclus(file=inputfile,
dbic=inputChans,
dboc=[chansCount + 1],
numclus=classesNumber,
maxiter=iterations,
movethrs=moveThresh)
finally:
enableDefaultReport('term') # this will close the report file
flag1 = time.time()
print "Classification complete! Time elapsed: " + str(
flag1 - start) + " seconds"
print ""
def kMeansClustering(fileInput, inputChannels,outputChannel):
mask = [] # Area or window in the input image to process
numberOfClasses = [5] # Number of classes
seedfile = ""
maximumIterations = [5] # Number of Iterations for calculating the mean positions
movethrs = [0.01] # Movement threshold
siggen = ""
backval = [] # Background gray level value to be ignored during classification
nsam = [] # Number of samples to collect on which to perform the iterative clustering
try:
kclus(fileInput, inputChannels, outputChannel, mask, numberOfClasses, seedfile, maximumIterations, movethrs, siggen, backval, nsam)
print "K means clustering finished"
except PCIException, e:
print e
def classification(self, file, alg):
if (alg == 'kmeans'):
print("Your selection is K-means Classification")
else:
print("Your selection is Iso-Cluster Classification")
report = os.getcwd() + "\\Data\\Result\\Report.txt"
if os.path.isfile(report):
print("Deleting old Report")
os.remove(report)
with datasource.open_dataset(file) as dataset:
cc = dataset.chan_count
print("Available Channels: ")
print(cc)
if (alg == 'kmeans'):
file = file
dbic = list(range(1, cc + 1, 1)) # input channels
dboc = [cc + 1] # output channel
print(dboc)
mask = [] # process entire image
numclus = [5] # requested number of clusters
seedfile = '' # automatically generate seeds
maxiter = [20] # no more than 20 iterations
movethrs = [0.01]
print("The file is: " + root.filename)
## enableDefaultReport(prev_report)
pcimod(file=file, pciop='add', pcival=[1, 0, 0, 0, 0, 0])
Report.clear()
enableDefaultReport(report)
print("Applying K-Means Classification")
kcluster = kclus(file, dbic, dboc, mask, numclus, seedfile,
maxiter, movethrs)
print("K-Means Classification is successful")
else:
file = file # input file
dbic = list(range(1, cc + 1, 1)) # input channels
dboc = [cc + 1] # output channel
print(dboc)
mask = [] # process entire image
numclus = [5] # requested number of clusters
maxclus = [7] # at most 20 clusters
minclus = [5] # at least 5 clusters
seedfile = '' # automatically generate seeds
maxiter = [5] # no more than 20 iterations
movethrs = [0.01]
siggen = "NO" # no signature generation
samprm = [5]
stdv = [10.0]
lump = [1.0]
maxpair = [5] # no more than 5 cluster center pairs
# clumped in one iteration
backval = [] # no background value
nsam = [] # default number of samples
print("The file is: " + root.filename)
## enableDefaultReport(prev_report)
pcimod(file=file, pciop='add', pcival=[1, 0, 0, 0, 0, 0])
Report.clear()
enableDefaultReport(report)
print("Applying Iso-cluster Classification")
isoclus( file, dbic, dboc, mask, numclus, maxclus, minclus, seedfile, maxiter,\
movethrs, siggen, samprm, stdv, lump, maxpair, backval, nsam )
print("Iso-Cluster Classification is successful")
pcimod(file=file, pciop='add', pcival=[1])
## root.status.set("Running fmo")
print("Applying MOD filter")
kfmo = fmo(file=file, dbic=[cc + 1], dboc=[cc + 2])
print("MOD filter is successfully applied")
pcimod(file=file, pciop='add', pcival=[1])
## root.status.set("Running sieve")
print("Applying SIEVE filter")
ksieve = sieve(file=file, dbic=[cc + 1], dboc=[cc + 3], sthresh=[32])
print("SIEVE filter is successfully applied")
## Split and insert directory name
split_file = file.split("/")
split_file.insert(-1, "Result")
join_result = "\\".join(split_file)
## Split by .
dot_result = join_result.split(".")
## print(dot_result)
## print("dot result")
## join .shp at last
shp_result = dot_result[0] + ".shp"
current_file_path, ext = os.path.splitext(str(file))
current_file = current_file_path.split("/")[-1]
## print(current_file)
## print("current file")
## Delete previously generated shapefiles
prev_files = glob.glob(os.getcwd() + "\\Data\\Result\\" + "*")
## print(prev_files)
## print("previous files")
for prev_file in prev_files:
prev_file_name, format = os.path.splitext(
str(os.path.basename(prev_file)))
## print(prev_file_name)
## print(" prev file names")
if (current_file in prev_file_name):
print("DELETING: " + str(prev_file))
os.remove(prev_file)
print("Successfully DELETED: " + str(prev_file))
## if(os.path.exists("../Data/Result/"+dot_result[0])):
## print("inside if ......")
## os.remove(shp_result)
## print("shp is deleted...")
print("Exporting to shapefile")
ras2poly(fili=file,
dbic=[cc + 3],
filo=shp_result,
smoothv="Yes",
ftype="SHP",
foptions="")
print("Shapefile is successfully created")
# Execution
qprint('Press any key to continue...')
input()
qprint('\n\n===== (>_<) Starting Execution (>_<) =====\n\n')
# KCLUS
from pci.kclus import kclus
qprintl('[1 of 5] KCLUS\tExecuting K-Means...')
kclus(
file=dataset.name, # Input file name
dbic=inputChannels, # Input channels
dboc=[kClusChannel], # Output channel
numclus=[5], # Number of output cluster classes
maxiter=[5] # Maximum iterations
)
updateChannelName(kClusChannel, KCLUS_RESULT)
qprintl('[1 of 5] KCLUS\tExecuting K-Means...OK \\(^o^)/')
# FMO
from pci.fmo import fmo
qprint('[2 of 5] FMO\tApplying Mode Filter...')
fmo(
file=dataset.name, # Input file
dbic=[kClusChannel], # Input channel
dboc=[fmoChannel], # Output channel
for i in range(1, dataset.chan_count + 1):
qprintl(' Channel {}: {}'.format(
i, auxiliaryData.get_chan_description(i)))
else:
qprintl(' No')
# KCLUS
# kclus: Unsupervised Classification with K-Means
from pci.kclus import kclus
qprintl('Executing K-Means...')
kclus(
file=dataset.name, # Input file name
dbic=list(range(1, 7)), # Input channels
dboc=[7], # Output channel
numclus=[5], # Number of output cluster classes
maxiter=[5] # Maximum iterations
)
auxiliaryData.set_chan_description('K-Means', 7)
qprintl('Executing K-Means: OK \\(^o^)/')
# FMO
# fmo: Mode Filter
from pci.fmo import fmo
qprint('Applying Mode Filter...')
fmo(
file=dataset.name, # Input file
dbic=[7], # Input channel
def classification(image):
# Set timer
start = time.time()
# Define output file name
outputFile = "GH_classPolygons.shp"
output = outputFolder + "\\" + outputFile
# get image statistics to extract channel number
print "Detecting current channels..."
with ds.open_dataset(image) as dataset:
chansCount = dataset.chan_count
print str(chansCount) + " channels detected"
print ""
# Whipe previously created channels (only use if you want a fresh clean input file and don't need previously created channels)
chansDel = range(7, chansCount + 1)
if chansCount > 6: # Adjust depending on image bands
pcimod(file=image, pciop='del', pcival=chansDel)
print str(len(chansDel)) + " previously created channels deleted"
# Whipe previously created shape files in output folder
files = glob.glob(outputFolder + "\\" + '*')
for f in files:
os.remove(f)
print "Previous output files deleted"
print ""
# Count input channels and create input channel list
with ds.open_dataset(image) as dataset:
inputChansCount = dataset.chan_count
inputChans = range(1, inputChansCount + 1)
print inputChans
# Add 3 channels to the image for storing algorithm output
print "Adding three 8-bit channels to image..."
pcimod(file=image, pciop='add', pcival=[3, 0, 0, 0, 0, 0])
print "Three 8-bit channels added"
print ""
# Run k-means cluster algorithm
classesNumber = [8] # define number of classes
iterations = [10] # define number iterations
moveThresh = [0.01] # define move threshhold
print "Running unsupervized k-means classification..."
print "Creating " + str(classesNumber) + " classes, applying " + str(
iterations) + " iterations at a move-threshhold of " + str(moveThresh)
kclus(file=image,
dbic=inputChans,
dboc=[7],
numclus=classesNumber,
maxiter=iterations,
movethrs=moveThresh)
flag1 = time.time()
print "Classification complete! Time ellapsed: " + str(flag1 -
start) + " seconds"
print ""
# Run mode filter
print "Running Mode Filter..."
fmo(file=image, dbic=[7], dboc=[8], thinline="OFF", flsz=[3, 3])
flag2 = time.time()
print "Filtering complete! Time ellapsed: " + str(flag2 -
start) + " seconds"
print ""
# Run sieve
print "Applying sieve..."
sieve(file=image, dbic=[8], dboc=[9], sthresh=[16])
flag3 = time.time()
print "Sieve complete! Time ellapsed: " + str(flag3 - start) + " seconds"
print ""
# Create vector ploygons and export as shape file
print "Creating polygons..."
ras2poly(fili=image, dbic=[9], filo=output, ftype="SHP")
flag4 = time.time()
print "Polygons created! Time ellapsed: " + str(flag4 - start) + " seconds"
print ""
print "Exporting as shape file..."
end = time.time()
print "Total time ellapsed: " + str(end - start) + " seconds"