def applyKmeansMasks(image_input, mask_samples_macro_input_list, image_samples_merged_output, proposal_table_output, micro_samples_images_output_list, centroids_files_output_list, macroclass_sampling_list, macroclass_labels_list, no_data_value, path_time_log, kmeans_param_maximum_iterations=200, kmeans_param_training_set_size_weight=1, kmeans_param_minimum_training_set_size=-1, rate_clean_micro_class=0.0, rand_otb=0, ram_otb=0, number_of_actives_pixels_threshold=200, extension_raster=".tif", save_results_intermediate=False, overwrite=True):
# Mise à jour du Log
starting_event = "applyKmeansMasks() : Kmeans and mask starting : "
timeLine(path_time_log,starting_event)
print(endC)
print(cyan + "applyKmeansMasks() : " + bold + green + "## START : SUBSAMPLING OF " + str(macroclass_labels_list) + endC)
print(endC)
if debug >= 2:
print(cyan + "applyKmeansMasks() : variables dans la fonction" + endC)
print(cyan + "applyKmeansMasks() : " + endC + "image_input : " + str(image_input) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "image_samples_merged_output : " + str(image_samples_merged_output) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "proposal_table_output : " + str(proposal_table_output) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "mask_samples_macro_input_list : " + str(mask_samples_macro_input_list) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "micro_samples_images_output_list : " + str(micro_samples_images_output_list) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "centroids_files_output_list : " + str(centroids_files_output_list) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "macroclass_sampling_list : " + str(macroclass_sampling_list) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "macroclass_labels_list : " + str(macroclass_labels_list) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "kmeans_param_maximum_iterations : " + str(kmeans_param_maximum_iterations) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "kmeans_param_training_set_size_weight : " + str(kmeans_param_training_set_size_weight) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "kmeans_param_minimum_training_set_size : " + str(kmeans_param_minimum_training_set_size) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "rate_clean_micro_class : " + str(rate_clean_micro_class))
print(cyan + "applyKmeansMasks() : " + endC + "no_data_value : " + str(no_data_value) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "rand_otb : " + str(rand_otb) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "ram_otb : " + str(ram_otb) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "number_of_actives_pixels_threshold : " + str(number_of_actives_pixels_threshold))
print(cyan + "applyKmeansMasks() : " + endC + "extension_raster : " + str(extension_raster) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "save_results_intermediate : " + str(save_results_intermediate) + endC)
print(cyan + "applyKmeansMasks() : " + endC + "overwrite : " + str(overwrite) + endC)
# constantes
HEADER_TABLEAU_MODIF = "MICROCLASSE;TRAITEMENT\n"
CODAGE_16B = "uint16"
CODAGE_8B = "uint8"
EXT_XML = ".xml"
SUFFIX_MASK_CLEAN = "_clean"
SUFFIX_SAMPLE_MICRO = "_sample_micro"
SUFFIX_STATISTICS = "_statistics"
SUFFIX_CENTROID = "_centroid"
SUFFIX_MASK_TEMP = "_tmp"
# Creation des fichiers temporaires de sortie si ils ne sont pas spécifier
#-------------------------------------------------------------------------
length_mask = len(mask_samples_macro_input_list)
images_mask_cleaned_list = []
temporary_files_list = []
micro_samples_images_list = []
centroids_files_list = []
repertory_output_tmp_list = []
if image_samples_merged_output != "" :
repertory_base_output = os.path.dirname(image_samples_merged_output)
filename = os.path.splitext(os.path.basename(image_samples_merged_output))[0]
else :
repertory_base_output = os.path.dirname(micro_samples_images_output_list[0])
filename = os.path.splitext(os.path.basename(micro_samples_images_output_list[0]))[0]
file_statistic_points = repertory_base_output + os.sep + filename + SUFFIX_STATISTICS + EXT_XML
for macroclass_id in range(length_mask):
repertory_output = repertory_base_output + os.sep + str(macroclass_labels_list[macroclass_id])
if not os.path.isdir(repertory_output):
os.makedirs(repertory_output)
repertory_output_tmp_list.append(repertory_output)
samples_image_input = mask_samples_macro_input_list[macroclass_id]
filename = os.path.splitext(os.path.basename(samples_image_input))[0]
image_mask_cleaned = repertory_output + os.sep + filename + SUFFIX_MASK_CLEAN + extension_raster
images_mask_cleaned_list.append(image_mask_cleaned)
image_tmp = repertory_output + os.sep + filename + SUFFIX_MASK_TEMP + extension_raster
temporary_files_list.append(image_tmp)
if micro_samples_images_output_list == [] :
micro_samples_image = repertory_output + os.sep + filename + SUFFIX_SAMPLE_MICRO + extension_raster
else :
micro_samples_image = micro_samples_images_output_list[macroclass_id]
micro_samples_images_list.append(micro_samples_image)
if centroids_files_output_list == [] :
centroids_file = repertory_output + os.sep + filename + SUFFIX_CENTROID + extension_raster
else :
centroids_file = centroids_files_output_list[macroclass_id]
centroids_files_list.append(centroids_file)
# Nettoyage des pixels superposés sur plusieurs images
#-----------------------------------------------------
if length_mask > 1:
image_name = os.path.splitext(os.path.basename(image_input))[0]
deletePixelsSuperpositionMasks(mask_samples_macro_input_list, images_mask_cleaned_list, image_name, CODAGE_8B)
else:
images_mask_cleaned_list = mask_samples_macro_input_list
# Execution du kmeans pour chaque macroclasse
#--------------------------------------------
# Initialisation de la liste pour le multi-threading
thread_list = []
for macroclass_id in range(length_mask):
mask_sample_input = images_mask_cleaned_list[macroclass_id]
micro_samples_image = micro_samples_images_list[macroclass_id]
image_tmp = temporary_files_list[macroclass_id]
centroids_file = centroids_files_list[macroclass_id]
check = os.path.isfile(micro_samples_image)
if check and not overwrite : # Si un fichier de sortie avec le même nom existe déjà, et si l'option ecrasement est à false, alors passe à la classification suivante
print(cyan + "applyKmeansMasks() : " + bold + yellow + "Computing kmeans from %s with %s already done : no actualisation" % (image_input, mask_sample_input) + endC)
else: # Si non, on applique un kmeans
if check :
removeFile(micro_samples_image) # Suppression de l'éventuel fichier existant
print(cyan + "applyKmeansMasks() : " + bold + green + "Computing kmeans from %s with %s ; output image is %s" %(image_input, mask_sample_input,micro_samples_image) + endC)
# Obtention du nombre de microclasses
number_of_classes = macroclass_sampling_list[macroclass_id] # Nombre de microclasses
label = macroclass_labels_list[macroclass_id] # Label de la macroclasse Ex : 11000
# Gestion du multi threading pour l'appel du calcul du kmeans
thread = threading.Thread(target=computeKmeans, args=(image_input, mask_sample_input, image_tmp, micro_samples_image, centroids_file, label, number_of_classes, macroclass_id, number_of_actives_pixels_threshold, kmeans_param_minimum_training_set_size, kmeans_param_maximum_iterations, length_mask, no_data_value, rand_otb, int(ram_otb/length_mask), CODAGE_8B, CODAGE_16B, save_results_intermediate, overwrite))
thread.start()
thread_list.append(thread)
# Start Kmeans all macro classes
try:
for thread in thread_list:
thread.join()
except:
print(cyan + "applyKmeansMasks() : " + bold + red + "applyKmeansMasks() : " + endC + "Erreur lors du calcul du kmeans : impossible de demarrer le thread" + endC, file=sys.stderr)
# Fusion des echantillons micro
#------------------------------
if image_samples_merged_output != "" :
mergeListRaster(micro_samples_images_list, image_samples_merged_output, CODAGE_16B)
updateReferenceProjection(image_input, image_samples_merged_output)
# Creation de la table de proposition et le fichier statistique
#--------------------------------------------------------------
if proposal_table_output != "" :
suppress_micro_class_list = []
info_micoclass_nbpoints_dico = {}
nb_points_total = 0
nb_points_medium = 0
# Liste des identifants des micro classes disponibles
id_micro_list = identifyPixelValues(image_samples_merged_output)
if 0 in id_micro_list :
id_micro_list.remove(0)
nb_micr_class = len(id_micro_list)
# Pour toutes les micro classes
for id_micro in id_micro_list :
nb_pixels = countPixelsOfValue(image_samples_merged_output, id_micro)
info_micoclass_nbpoints_dico[id_micro] = nb_pixels
nb_points_total += nb_pixels
# Valeur moyenne de nombre de points
if nb_micr_class != 0 :
nb_points_medium = int(nb_points_total / nb_micr_class)
nb_points_min = int((nb_points_medium * rate_clean_micro_class) / 100)
# Identifier les micro classes trop petites
if debug >= 4:
print("rate_clean_micro_class = " + str(rate_clean_micro_class))
print("nb_points_medium = " + str(nb_points_medium))
print("nb_points_min = " + str(nb_points_min))
# Preparation du fichier statistique
writeTextFile(file_statistic_points, '<?xml version="1.0" ?>\n')
appendTextFileCR(file_statistic_points, '<GeneralStatistics>')
appendTextFileCR(file_statistic_points, ' <Statistic name="pointsPerClassRaw">')
for micro_class_id in info_micoclass_nbpoints_dico :
nb_points = info_micoclass_nbpoints_dico[micro_class_id]
if debug >= 4:
print("micro_class_id = " + str(micro_class_id) + ", nb_points = " + str(nb_points))
appendTextFileCR(file_statistic_points, ' <StatisticPoints class="%d" value="%d" />' %(micro_class_id, nb_points))
if nb_points < nb_points_min :
# Micro_class à proposer en effacement
suppress_micro_class_list.append(micro_class_id)
# Fin du fichier statistique
appendTextFileCR(file_statistic_points, ' </Statistic>')
appendTextFileCR(file_statistic_points, '</GeneralStatistics>')
# Test si ecrassement de la table précédemment créée
check = os.path.isfile(proposal_table_output)
if check and not overwrite :
print(cyan + "applyKmeansMasks() : " + bold + yellow + "Modifier table already exists." + '\n' + endC)
else:
# Tenter de supprimer le fichier
try:
removeFile(proposal_table_output)
except Exception:
pass # Ignore l'exception levee si le fichier n'existe pas (et ne peut donc pas être supprime)
# lister les micro classes à supprimer
text_output = HEADER_TABLEAU_MODIF
for micro_class_del in suppress_micro_class_list:
text_output += "%d;-1\n" %(micro_class_del)
# Ecriture du fichier proposition de réaffectation
writeTextFile(proposal_table_output, text_output)
# Suppresions fichiers intermediaires inutiles
#---------------------------------------------
if not save_results_intermediate:
for macroclass_id in range(length_mask):
if (os.path.isfile(temporary_files_list[macroclass_id])) :
removeFile(temporary_files_list[macroclass_id])
if (length_mask > 1) and (os.path.isfile(images_mask_cleaned_list[macroclass_id])) :
removeFile(images_mask_cleaned_list[macroclass_id])
if (micro_samples_images_output_list == []) and (os.path.isfile(micro_samples_images_list[macroclass_id])) :
removeFile(micro_samples_images_list[macroclass_id])
if (centroids_files_output_list == []) and (os.path.isfile(centroids_files_list[macroclass_id])) :
removeFile(centroids_files_list[macroclass_id])
if os.path.isdir(repertory_output_tmp_list[macroclass_id]) :
removeDir(repertory_output_tmp_list[macroclass_id])
print(cyan + "applyKmeansMasks() : " + bold + green + "## END : KMEANS CLASSIFICATION" + endC)
print(endC)
# Mise à jour du Log
ending_event = "applyKmeansMasks() : Kmeans and mask ending : "
timeLine(path_time_log,ending_event)
return
def computeKmeans(image_input, mask_sample_input, image_output, micro_samples_image_out, centroids_file_output, label, number_of_classes, macroclass_id, number_of_actives_pixels_threshold, kmeans_param_minimum_training_set_size, kmeans_param_maximum_iterations, length_mask, no_data_value, rand_otb, ram_otb, codage_8b, codage_16b, save_results_intermediate=False, overwrite=True):
# ETAPE 0 : PREPARATION
if debug >= 4:
print(cyan + "computeKmeans() : " + endC + "image : " + str(image_input) + endC)
print(cyan + "computeKmeans() : " + endC + "label : " + str(label) + endC)
print(cyan + "computeKmeans() : " + endC + "number_of_classes : " + str(number_of_classes) + endC)
print(cyan + "computeKmeans() : " + endC + "macroclass_id : " + str(macroclass_id) + endC)
print(cyan + "computeKmeans() : " + endC + "mask_sample_input : " + str(mask_sample_input) + endC)
print(cyan + "computeKmeans() : " + endC + "image_output : " + str(image_output) + endC)
print(cyan + "computeKmeans() : " + endC + "micro_samples_image_out : " + str(micro_samples_image_out) + endC)
print(cyan + "computeKmeans() : " + endC + "centroids_file_output : " + str(centroids_file_output) + endC)
print(cyan + "computeKmeans() : " + endC + "kmeans_param_minimum_training_set_size : " + str(training_set_size) + endC)
print(cyan + "computeKmeans() : " + endC + "number_of_classes : " + str(number_of_classes) + endC)
print(cyan + "computeKmeans() : " + endC + "kmeans_param_maximum_iterations : " + str(kmeans_param_maximum_iterations) + endC)
print(cyan + "computeKmeans() : " + endC + "no_data_value : " + str(no_data_value) + endC)
print(cyan + "computeKmeans() : " + endC + "rand_otb : " + str(rand_otb) + endC)
print(cyan + "computeKmeans() : " + endC + "ram_otb : " + str(ram_otb) + endC)
print(cyan + "computeKmeans() : " + endC + "save_results_intermediate : " + str(save_results_intermediate) + endC)
print(cyan + "computeKmeans() : " + endC + "overwrite : " + str(overwrite) + endC)
# kmeans_param_minimum_training_set_size ?
if kmeans_param_minimum_training_set_size == -1:
# Le nombre de pixels d'apprentissage correspond au nombre de pixels à 1 du masque "mask_sample_input"
training_set_size = countPixelsOfValue(mask_sample_input, 1)
else :
training_set_size = kmeans_param_minimum_training_set_size
# ETAPE 1 : CLASSIFICATION NON SUPERVISEE
# Cas où il y a moins de pixels disponibles pour effectuer le kmeans que le seuil
if training_set_size < (number_of_classes * number_of_actives_pixels_threshold) :
print(cyan + "computeKmeans() : " + bold + yellow + "MACROCLASSE %s / %s (%s): Nombre insuffisant de pixels disponibles pour appliquer le kmeans : %s sur %s requis au minimum " %(macroclass_id, length_mask, label, training_set_size, number_of_classes * number_of_actives_pixels_threshold) + endC)
if debug >= 2:
print(cyan + "computeKmeans() : " + bold + yellow + "MACROCLASSE %s / %s : SOUS ECHANTILLONAGE NON APPLIQUE A LA CLASSE %s" %(macroclass_id + 1, length_mask, label) + endC)
print(cyan + "computeKmeans() : " + bold + yellow + "MACROCLASSE %s / %s : COPIE DE %s A %s" %(macroclass_id + 1, length_mask, mask_sample_input, micro_samples_image_out) + endC + "\n")
# Recopie de fichier d'entré mask
shutil.copy2(mask_sample_input, image_output)
else: # Cas où il y a suffisamment de pixels pour effectuer le kmeans
if centroids_file_output != None: # Distinction des cas avec et sans coordonnees des centroides
command = "otbcli_KMeansClassification -in %s -out %s %s -vm %s -ts %s -nc %s -maxit %s" %(image_input, image_output, codage_8b, mask_sample_input, training_set_size, number_of_classes, kmeans_param_maximum_iterations)
if IS_VERSION_UPPER_OTB_7_0 :
command += " -centroids.out %s " %(centroids_file_output)
else :
command += " -outmeans %s " %(centroids_file_output)
if no_data_value != 0:
command += " -nodatalabel %d" %(no_data_value)
if rand_otb > 0:
command += " -rand %d" %(rand_otb)
if ram_otb > 0:
command += " -ram %d" %(ram_otb)
if debug >=3:
print(cyan + "computeKmeans() : " + bold + green + "MACROCLASSE %s / %s (%s): ETAPE 1 : Nombre suffisant de pixels disponibles %s sur %s requis au minimum " %(macroclass_id + 1, length_mask, label, training_set_size, number_of_classes * number_of_actives_pixels_threshold) + endC)
if debug >=2:
print(cyan + "computeKmeans() : " + bold + green + "MACROCLASSE %s / %s (%s): ETAPE 1 : Computing kmeans from %s " %(macroclass_id + 1, length_mask, label, image_input) + endC)
print(cyan + "computeKmeans() : " + bold + green + "Mask : %s " %(mask_sample_input) + endC)
print(cyan + "computeKmeans() : " + bold + green + "Output image : %s" %(micro_samples_image_out) + endC)
print(command)
exitCode = os.system(command)
if exitCode != 0:
raise NameError(cyan + "computeKmeans() : " + bold + red + "An error occured during otbcli_KMeansClassification command. See error message above." + endC)
else :
command = "otbcli_KMeansClassification -in %s -out %s %s -vm %s -ts %s -nc %s -maxit %s" %(image_input, image_output, codage_8b, mask_sample_input, str(training_set_size), str(number_of_classes), str(kmeans_param_maximum_iterations))
if no_data_value != 0:
command += " -nodatalabel %d" %(no_data_value)
if rand_otb > 0:
command += " -rand %d" %(rand_otb)
if ram_otb > 0:
command += " -ram %d" %(ram_otb)
if debug >=2:
print(cyan + "computeKmeans() : " + bold + green + "MACROCLASSE %s / %s (%s): ETAPE 1 : Computing kmeans from %s with %s ; output image is %s" %(macroclass_id + 1, length_mask,label,image_input, mask_sample_input,micro_samples_image_out) + endC)
print(command)
exitCode = os.system(command)
if exitCode != 0:
raise NameError(cyan + "computeKmeans() : " + bold + red + "An error occured during otbcli_KMeansClassification command. See error message above." + endC)
# ETAPE 2 : GESTION DU SYSTEME DE PROJECTION
if debug >=2:
print(cyan + "computeKmeans() : " + bold + green + "MACROCLASSE %s / %s (%s): ETAPE 2 : GESTION DU SYSTEME DE PROJECTION" %(macroclass_id + 1, length_mask, label) + endC)
updateReferenceProjection (image_input, image_output)
# ETAPE 3 : APPLICATION DU MASQUE ET LABELLISATION EN MICROCLASSES
expression = "\"(im1b1+%s)*im2b1\"" %(str(label)) # Expression qui passe à 0 les pixels masqués et qui labelise à macroclass_label+classe du computeKmeans
command = "otbcli_BandMath -il %s %s -out %s %s -exp %s" %(image_output, mask_sample_input, micro_samples_image_out, codage_16b, expression)
if ram_otb > 0:
command += " -ram %d" %(ram_otb)
if debug >=2:
print(cyan + "computeKmeans() : " + bold + green + "MACROCLASSE %s / %s (%s): ETAPE 3 : APPLICATION DU MASQUE ET LABELLISATION EN MICROCLASSES" %(macroclass_id + 1, length_mask,label) + endC)
print(command)
exitCode = os.system(command)
if exitCode != 0:
print(command)
raise NameError(cyan + "computeKmeans() : " + bold + red + "An error occured during otbcli_BandMath command. See error message above." + endC)
return
def createMnh(image_mns_input, image_mnt_input, image_threshold_input, vector_emprise_input, image_mnh_output, automatic, bd_road_vector_input_list, bd_road_buff_list, sql_road_expression_list, bd_build_vector_input_list, height_bias, threshold_bd_value, threshold_delta_h, mode_interpolation, method_interpolation, interpolation_bco_radius, simplify_vector_param, epsg, no_data_value, ram_otb, path_time_log, format_raster='GTiff', format_vector='ESRI Shapefile', extension_raster=".tif", extension_vector=".shp", save_results_intermediate=False, overwrite=True):
# Mise à jour du Log
starting_event = "createMnh() : MNH creation starting : "
timeLine(path_time_log,starting_event)
print(endC)
print(bold + green + "## START : MNH CREATION" + endC)
print(endC)
if debug >= 2:
print(bold + green + "createMnh() : Variables dans la fonction" + endC)
print(cyan + "createMnh() : " + endC + "image_mns_input : " + str(image_mns_input) + endC)
print(cyan + "createMnh() : " + endC + "image_mnt_input : " + str(image_mnt_input) + endC)
print(cyan + "createMnh() : " + endC + "image_threshold_input : " + str(image_threshold_input) + endC)
print(cyan + "createMnh() : " + endC + "vector_emprise_input : " + str(vector_emprise_input) + endC)
print(cyan + "createMnh() : " + endC + "image_mnh_output : " + str(image_mnh_output) + endC)
print(cyan + "createMnh() : " + endC + "automatic : " + str(automatic) + endC)
print(cyan + "createMnh() : " + endC + "bd_road_vector_input_list : " + str(bd_road_vector_input_list) + endC)
print(cyan + "createMnh() : " + endC + "bd_road_buff_list : " + str(bd_road_buff_list) + endC)
print(cyan + "createMnh() : " + endC + "sql_road_expression_list : " + str(sql_road_expression_list) + endC)
print(cyan + "createMnh() : " + endC + "bd_build_vector_input_list : " + str(bd_build_vector_input_list) + endC)
print(cyan + "createMnh() : " + endC + "height_bias : " + str(height_bias) + endC)
print(cyan + "createMnh() : " + endC + "threshold_bd_value : " + str(threshold_bd_value) + endC)
print(cyan + "createMnh() : " + endC + "threshold_delta_h : " + str(threshold_delta_h) + endC)
print(cyan + "createMnh() : " + endC + "mode_interpolation : " + str(mode_interpolation) + endC)
print(cyan + "createMnh() : " + endC + "method_interpolation : " + str(method_interpolation) + endC)
print(cyan + "createMnh() : " + endC + "interpolation_bco_radius : " + str(interpolation_bco_radius) + endC)
print(cyan + "createMnh() : " + endC + "simplify_vector_param : " + str(simplify_vector_param) + endC)
print(cyan + "createMnh() : " + endC + "epsg : " + str(epsg) + endC)
print(cyan + "createMnh() : " + endC + "no_data_value : " + str(no_data_value) + endC)
print(cyan + "createMnh() : " + endC + "ram_otb : " + str(ram_otb) + endC)
print(cyan + "createMnh() : " + endC + "path_time_log : " + str(path_time_log) + endC)
print(cyan + "createMnh() : " + endC + "format_raster : " + str(format_raster) + endC)
print(cyan + "createMnh() : " + endC + "format_vector : " + str(format_vector) + endC)
print(cyan + "createMnh() : " + endC + "extension_raster : " + str(extension_raster) + endC)
print(cyan + "createMnh() : " + endC + "extension_vector : " + str(extension_vector) + endC)
print(cyan + "createMnh() : " + endC + "save_results_intermediate : " + str(save_results_intermediate) + endC)
print(cyan + "createMnh() : " + endC + "overwrite : " + str(overwrite) + endC)
# LES CONSTANTES
PRECISION = 0.0000001
CODAGE_8B = "uint8"
CODAGE_F = "float"
SUFFIX_CUT = "_cut"
SUFFIX_CLEAN = "_clean"
SUFFIX_SAMPLE = "_sample"
SUFFIX_MASK = "_mask"
SUFFIX_TMP = "_tmp"
SUFFIX_MNS = "_mns"
SUFFIX_MNT = "_mnt"
SUFFIX_ROAD = "_road"
SUFFIX_BUILD = "_build"
SUFFIX_RASTER = "_raster"
SUFFIX_VECTOR = "_vector"
# DEFINIR LES REPERTOIRES ET FICHIERS TEMPORAIRES
repertory_output = os.path.dirname(image_mnh_output)
basename_mnh = os.path.splitext(os.path.basename(image_mnh_output))[0]
sub_repertory_raster_temp = repertory_output + os.sep + basename_mnh + SUFFIX_RASTER + SUFFIX_TMP
sub_repertory_vector_temp = repertory_output + os.sep + basename_mnh + SUFFIX_VECTOR + SUFFIX_TMP
cleanTempData(sub_repertory_raster_temp)
cleanTempData(sub_repertory_vector_temp)
basename_vector_emprise = os.path.splitext(os.path.basename(vector_emprise_input))[0]
basename_mns_input = os.path.splitext(os.path.basename(image_mns_input))[0]
basename_mnt_input = os.path.splitext(os.path.basename(image_mnt_input))[0]
image_mnh_tmp = sub_repertory_raster_temp + os.sep + basename_mnh + SUFFIX_TMP + extension_raster
image_mnh_road = sub_repertory_raster_temp + os.sep + basename_mnh + SUFFIX_ROAD + extension_raster
vector_bd_bati_temp = sub_repertory_vector_temp + os.sep + basename_mnh + SUFFIX_BUILD + SUFFIX_TMP + extension_vector
vector_bd_bati = repertory_output + os.sep + basename_mnh + SUFFIX_BUILD + extension_vector
raster_bd_bati = sub_repertory_vector_temp + os.sep + basename_mnh + SUFFIX_BUILD + extension_raster
removeVectorFile(vector_bd_bati)
image_emprise_mnt_mask = sub_repertory_raster_temp + os.sep + basename_vector_emprise + SUFFIX_MNT + extension_raster
image_mnt_cut = sub_repertory_raster_temp + os.sep + basename_mnt_input + SUFFIX_CUT + extension_raster
image_mnt_clean = sub_repertory_raster_temp + os.sep + basename_mnt_input + SUFFIX_CLEAN + extension_raster
image_mnt_clean_sample = sub_repertory_raster_temp + os.sep + basename_mnt_input + SUFFIX_CLEAN + SUFFIX_SAMPLE + extension_raster
image_emprise_mns_mask = sub_repertory_raster_temp + os.sep + basename_vector_emprise + SUFFIX_MNS + extension_raster
image_mns_cut = sub_repertory_raster_temp + os.sep + basename_mns_input + SUFFIX_CUT + extension_raster
image_mns_clean = sub_repertory_raster_temp + os.sep + basename_mns_input + SUFFIX_CLEAN + extension_raster
vector_bd_road_temp = sub_repertory_vector_temp + os.sep + basename_mnh + SUFFIX_ROAD + SUFFIX_TMP + extension_vector
raster_bd_road_mask = sub_repertory_raster_temp + os.sep + basename_mnh + SUFFIX_ROAD + SUFFIX_MASK + extension_raster
if image_threshold_input != "" :
basename_threshold_input = os.path.splitext(os.path.basename(image_threshold_input))[0]
image_threshold_cut = sub_repertory_raster_temp + os.sep + basename_threshold_input + SUFFIX_CUT + extension_raster
image_threshold_mask = sub_repertory_raster_temp + os.sep + basename_threshold_input + SUFFIX_MASK + extension_raster
# VERIFICATION SI LE FICHIER DE SORTIE EXISTE DEJA
# Si un fichier de sortie avec le même nom existe déjà, et si l'option ecrasement est à false, alors on ne fait rien
check = os.path.isfile(image_mnh_output)
if check and not overwrite:
print(bold + yellow + "createMnh() : " + endC + "Create mnh %s from %s and %s already done : no actualisation" % (image_mnh_output, image_mns_input, image_mnt_input) + endC)
# Si non, ou si la fonction ecrasement est désative, alors on le calcule
else:
if check:
try: # Suppression de l'éventuel fichier existant
removeFile(image_mnh_output)
except Exception:
pass # Si le fichier ne peut pas être supprimé, on suppose qu'il n'existe pas et on passe à la suite
# DECOUPAGE DES FICHIERS MS ET MNT D'ENTREE PAR LE FICHIER D'EMPRISE
if debug >= 3:
print(bold + green + "createMnh() : " + endC + "Decoupage selon l'emprise des fichiers %s et %s " %(image_mns_input, image_mnt_input) + endC)
# Fonction de découpe du mns
if not cutImageByVector(vector_emprise_input, image_mns_input, image_mns_cut, None, None, no_data_value, epsg, format_raster, format_vector) :
raise NameError (cyan + "createMnh() : " + bold + red + "!!! Une erreur c'est produite au cours du decoupage de l'image : " + image_mns_input + ". Voir message d'erreur." + endC)
# Fonction de découpe du mnt
if not cutImageByVector(vector_emprise_input, image_mnt_input, image_mnt_cut, None, None, no_data_value, epsg, format_raster, format_vector) :
raise NameError (cyan + "createMnh() : " + bold + red + "!!! Une erreur c'est produite au cours du decoupage de l'image : " + image_mnt_input + ". Voir message d'erreur." + endC)
if debug >= 3:
print(bold + green + "createMnh() : " + endC + "Decoupage des fichiers %s et %s complet" %(image_mns_cut, image_mnt_cut) + endC)
# REBOUCHAGE DES TROUS DANS LE MNT D'ENTREE SI NECESSAIRE
nodata_mnt = getNodataValueImage(image_mnt_cut)
pixelNodataCount = countPixelsOfValue(image_mnt_cut, nodata_mnt)
if pixelNodataCount > 0 :
if debug >= 3:
print(bold + green + "createMnh() : " + endC + "Fill the holes MNT for %s" %(image_mnt_cut) + endC)
# Rasterisation du vecteur d'emprise pour creer un masque pour boucher les trous du MNT
rasterizeBinaryVector(vector_emprise_input, image_mnt_cut, image_emprise_mnt_mask, 1, CODAGE_8B)
# Utilisation de SAGA pour boucher les trous
fillNodata(image_mnt_cut, image_emprise_mnt_mask, image_mnt_clean, save_results_intermediate)
if debug >= 3:
print(bold + green + "createMnh() : " + endC + "Fill the holes MNT to %s completed" %(image_mnt_clean) + endC)
else :
image_mnt_clean = image_mnt_cut
if debug >= 3:
print(bold + green + "\ncreateMnh() : " + endC + "Fill the holes not necessary MNT for %s" %(image_mnt_cut) + endC)
# REBOUCHAGE DES TROUS DANS LE MNS D'ENTREE SI NECESSAIRE
nodata_mns = getNodataValueImage(image_mns_cut)
pixelNodataCount = countPixelsOfValue(image_mns_cut, nodata_mns)
if pixelNodataCount > 0 :
if debug >= 3:
print(bold + green + "createMnh() : " + endC + "Fill the holes MNS for %s" %(image_mns_cut) + endC)
# Rasterisation du vecteur d'emprise pour creer un masque pour boucher les trous du MNS
rasterizeBinaryVector(vector_emprise_input, image_mns_cut, image_emprise_mns_mask, 1, CODAGE_8B)
# Utilisation de SAGA pour boucher les trous
fillNodata(image_mns_cut, image_emprise_mns_mask, image_mns_clean, save_results_intermediate)
if debug >= 3:
print(bold + green + "\ncreateMnh() : " + endC + "Fill the holes MNS to %s completed" %(image_mns_clean) + endC)
else :
image_mns_clean = image_mns_cut
if debug >= 3:
print(bold + green + "createMnh() : " + endC + "Fill the holes not necessary MNS for %s" %(image_mns_cut) + endC)
# CALLER LE FICHIER MNT AU FORMAT DU FICHIER MNS
# Commande de mise en place de la geométrie re-echantionage
command = "otbcli_Superimpose -inr " + image_mns_clean + " -inm " + image_mnt_clean + " -mode " + mode_interpolation + " -interpolator " + method_interpolation + " -out " + image_mnt_clean_sample
if method_interpolation.lower() == 'bco' :
command += " -interpolator.bco.radius " + str(interpolation_bco_radius)
if ram_otb > 0:
command += " -ram %d" %(ram_otb)
if debug >= 3:
print(cyan + "createMnh() : " + bold + green + "Réechantillonage du fichier %s par rapport à la reference %s" %(image_mnt_clean, image_mns_clean) + endC)
print(command)
exit_code = os.system(command)
if exit_code != 0:
print(command)
raise NameError (cyan + "createMnh() : " + bold + red + "!!! Une erreur c'est produite au cours du superimpose de l'image : " + image_mnt_input + ". Voir message d'erreur." + endC)
# INCRUSTATION DANS LE MNH DES DONNEES VECTEURS ROUTES
if debug >= 3:
print(bold + green + "createMnh() : " + endC + "Use BD road to clean MNH" + endC)
# Creation d'un masque de filtrage des donnes routes (exemple : le NDVI)
if image_threshold_input != "" :
if not cutImageByVector(vector_emprise_input, image_threshold_input, image_threshold_cut, None, None, no_data_value, epsg, format_raster, format_vector) :
raise NameError (cyan + "createMnh() : " + bold + red + "!!! Une erreur c'est produite au cours du decoupage de l'image : " + image_threshold_input + ". Voir message d'erreur." + endC)
createBinaryMask(image_threshold_cut, image_threshold_mask, threshold_bd_value, False, CODAGE_8B)
# Execution de la fonction createMacroSamples pour une image correspondant au données routes
if bd_road_vector_input_list != [] :
createMacroSamples(image_mns_clean, vector_emprise_input, vector_bd_road_temp, raster_bd_road_mask, bd_road_vector_input_list, bd_road_buff_list, sql_road_expression_list, path_time_log, basename_mnh, simplify_vector_param, format_vector, extension_vector, save_results_intermediate, overwrite)
if debug >= 3:
print(bold + green + "\ncreateMnh() : " + endC + "File raster from BD road is create %s" %(raster_bd_road_mask) + endC)
# CALCUL DU MNH
# Calcul par bandMath du MNH definir l'expression qui soustrait le MNT au MNS en introduisant le biais et en mettant les valeurs à 0 à une valeur approcher de 0.0000001
delta = ""
if height_bias > 0 :
delta = "+%s" %(str(height_bias))
elif height_bias < 0 :
delta = "-%s" %(str(abs(height_bias)))
else :
delta = ""
# Definition de l'expression
if bd_road_vector_input_list != [] :
if image_threshold_input != "" :
expression = "\"im3b1 > 0 and im4b1 > 0?%s:(im1b1-im2b1%s) > 0.0?im1b1-im2b1%s:%s\"" %(str(PRECISION), delta, delta, str(PRECISION))
command = "otbcli_BandMath -il %s %s %s %s -out %s %s -exp %s" %(image_mns_clean, image_mnt_clean_sample, raster_bd_road_mask, image_threshold_mask, image_mnh_tmp, CODAGE_F, expression)
else :
expression = "\"im3b1 > 0?%s:(im1b1-im2b1%s) > 0.0?im1b1-im2b1%s:%s\"" %(str(PRECISION), delta, delta, str(PRECISION))
command = "otbcli_BandMath -il %s %s %s -out %s %s -exp %s" %(image_mns_clean, image_mnt_clean_sample, raster_bd_road_mask, image_mnh_tmp, CODAGE_F, expression)
else :
expression = "\"(im1b1-im2b1%s) > 0.0?im1b1-im2b1%s:%s\"" %(delta, delta, str(PRECISION))
command = "otbcli_BandMath -il %s %s -out %s %s -exp %s" %(image_mns_clean, image_mnt_clean_sample, image_mnh_tmp, CODAGE_F, expression)
if ram_otb > 0:
command += " -ram %d" %(ram_otb)
if debug >= 3:
print(cyan + "createMnh() : " + bold + green + "Calcul du MNH %s difference du MNS : %s par le MNT :%s" %(image_mnh_tmp, image_mns_clean, image_mnt_clean_sample) + endC)
print(command)
exitCode = os.system(command)
if exitCode != 0:
print(command)
raise NameError(cyan + "createMnh() : " + bold + red + "An error occured during otbcli_BandMath command to compute MNH " + image_mnh_tmp + ". See error message above." + endC)
# DECOUPAGE DU MNH
if bd_build_vector_input_list == []:
image_mnh_road = image_mnh_output
if debug >= 3:
print(bold + green + "createMnh() : " + endC + "Decoupage selon l'emprise du fichier mnh %s " %(image_mnh_tmp) + endC)
# Fonction de découpe du mnh
if not cutImageByVector(vector_emprise_input, image_mnh_tmp, image_mnh_road, None, None, no_data_value, epsg, format_raster, format_vector) :
raise NameError (cyan + "createMnh() : " + bold + red + "!!! Une erreur c'est produite au cours du decoupage de l'image : " + image_mns_input + ". Voir message d'erreur." + endC)
if debug >= 3:
print(bold + green + "createMnh() : " + endC + "Decoupage du fichier mnh %s complet" %(image_mnh_road) + endC)
# INCRUSTATION DANS LE MNH DES DONNEES VECTEURS BATIS
# Si demander => liste de fichier vecteur bati passé en donnée d'entrée
if bd_build_vector_input_list != []:
# Découpage des vecteurs de bd bati exogenes avec l'emprise
vectors_build_cut_list = []
for vector_build_input in bd_build_vector_input_list :
vector_name = os.path.splitext(os.path.basename(vector_build_input))[0]
vector_build_cut = sub_repertory_vector_temp + os.sep + vector_name + SUFFIX_CUT + extension_vector
vectors_build_cut_list.append(vector_build_cut)
cutoutVectors(vector_emprise_input, bd_build_vector_input_list, vectors_build_cut_list, format_vector)
# Fusion des vecteurs batis découpés
fusionVectors (vectors_build_cut_list, vector_bd_bati_temp)
# Croisement vecteur rasteur entre le vecteur fusion des batis et le MNH créé precedement
statisticsVectorRaster(image_mnh_road, vector_bd_bati_temp, "", 1, False, False, True, ['PREC_PLANI','PREC_ALTI','ORIGIN_BAT','median','sum','std','unique','range'], [], {}, path_time_log, True, format_vector, save_results_intermediate, overwrite)
# Calcul de la colonne delta_H entre les hauteurs des batis et la hauteur moyenne du MNH sous le bati
COLUMN_ID = "ID"
COLUMN_H_BUILD = "HAUTEUR"
COLUMN_H_BUILD_MIN = "Z_MIN"
COLUMN_H_BUILD_MAX = "Z_MAX"
COLUMN_H_MNH = "mean"
COLUMN_H_MNH_MIN = "min"
COLUMN_H_MNH_MAX = "max"
COLUMN_H_DIFF = "H_diff"
field_type = ogr.OFTReal
field_value = 0.0
field_width = 20
field_precision = 2
attribute_name_dico = {}
attribute_name_dico[COLUMN_ID] = ogr.OFTString
attribute_name_dico[COLUMN_H_BUILD] = ogr.OFTReal
attribute_name_dico[COLUMN_H_MNH] = ogr.OFTReal
# Ajouter la nouvelle colonne H_diff
addNewFieldVector(vector_bd_bati_temp, COLUMN_H_DIFF, field_type, field_value, field_width, field_precision, format_vector)
# Recuperer les valeur de hauteur du bati et du mnt dans le vecteur
data_z_dico = getAttributeValues(vector_bd_bati_temp, None, None, attribute_name_dico, format_vector)
# Calculer la difference des Hauteur bati et mnt
field_new_values_dico = {}
for index in range(len(data_z_dico[COLUMN_ID])) :
index_polygon = data_z_dico[COLUMN_ID][index]
delta_h = abs(data_z_dico[COLUMN_H_BUILD][index] - data_z_dico[COLUMN_H_MNH][index])
field_new_values_dico[index_polygon] = {COLUMN_H_DIFF:delta_h}
# Mettre à jour la colonne H_diff dans le vecteur
setAttributeIndexValuesList(vector_bd_bati_temp, COLUMN_ID, field_new_values_dico, format_vector)
# Suppression de tous les polygones bati dons la valeur du delat H est inferieur à threshold_delta_h
column = "'%s, %s, %s, %s, %s, %s, %s, %s'"% (COLUMN_ID, COLUMN_H_BUILD, COLUMN_H_BUILD_MIN, COLUMN_H_BUILD_MAX, COLUMN_H_MNH, COLUMN_H_MNH_MIN, COLUMN_H_MNH_MAX, COLUMN_H_DIFF)
expression = "%s > %s" % (COLUMN_H_DIFF, threshold_delta_h)
filterSelectDataVector(vector_bd_bati_temp, vector_bd_bati, column, expression, overwrite, format_vector)
# Attention!!!! PAUSE pour trie et verification des polygones bati nom deja present dans le MNH ou non
if not automatic :
print(bold + blue + "Application MnhCreation => " + endC + "Vérification manuelle du vecteur bati %s pour ne concerver que les batis non présent dans le MNH courant %s" %(vector_bd_bati_temp, image_mnh_road) + endC)
input(bold + red + "Appuyez sur entree pour continuer le programme..." + endC)
# Creation du masque bati avec pour H la hauteur des batiments
rasterizeVector(vector_bd_bati, raster_bd_bati, image_mnh_road, COLUMN_H_BUILD)
# Fusion du mask des batis et du MNH temporaire
expression = "\"im1b1 > 0.0?im1b1:im2b1\""
command = "otbcli_BandMath -il %s %s -out %s %s -exp %s" %(raster_bd_bati, image_mnh_road, image_mnh_output, CODAGE_F, expression)
if ram_otb > 0:
command += " -ram %d" %(ram_otb)
if debug >= 3:
print(cyan + "createMnh() : " + bold + green + "Amelioration du MNH %s ajout des hauteurs des batis %s" %(image_mnh_road, raster_bd_bati) + endC)
print(command)
exitCode = os.system(command)
if exitCode != 0:
print(command)
raise NameError(cyan + "createMnh() : " + bold + red + "An error occured during otbcli_BandMath command to compute MNH Final" + image_mnh_output + ". See error message above." + endC)
# SUPPRESIONS FICHIERS INTERMEDIAIRES INUTILES
# Suppression des fichiers intermédiaires
if not save_results_intermediate :
if bd_build_vector_input_list != []:
removeFile(image_mnh_road)
removeFile(image_threshold_cut)
removeFile(image_threshold_mask)
removeFile(raster_bd_bati)
removeVectorFile(vector_bd_road_temp)
removeVectorFile(vector_bd_bati_temp)
removeVectorFile(vector_bd_bati) # A confirmer!!!
removeFile(raster_bd_road_mask)
removeFile(image_mnh_tmp)
deleteDir(sub_repertory_raster_temp)
deleteDir(sub_repertory_vector_temp)
print(endC)
print(bold + green + "## END : MNH CREATION" + endC)
print(endC)
# Mise à jour du Log
ending_event = "createMnh() : MNH creation ending : "
timeLine(path_time_log,ending_event)
return
def classRasterSubSampling(satellite_image_input, classified_image_input, image_output, table_reallocation, sub_sampling_number, no_data_value, path_time_log, rand_otb=0, ram_otb=0, number_of_actives_pixels_threshold=8000, extension_raster=".tif", save_results_intermediate=False, overwrite=True) :
# Mise à jour du Log
starting_event = "classRasterSubSampling() : Micro class subsampling on classification image starting : "
timeLine(path_time_log,starting_event)
if debug >= 3:
print(cyan + "classRasterSubSampling() : " + endC + "satellite_image_input : " + str(satellite_image_input) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "classified_image_input : " + str(classified_image_input) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "image_output : " + str(image_output) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "table_reallocation : " + str(table_reallocation) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "sub_sampling_number : " + str(sub_sampling_number) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "no_data_value : " + str(no_data_value) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "path_time_log : " + str(path_time_log) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "rand_otb : " + str(rand_otb) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "ram_otb : " + str(ram_otb) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "number_of_actives_pixels_threshold : " + str(number_of_actives_pixels_threshold) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "extension_raster : " + str(extension_raster) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "save_results_intermediate : " + str(save_results_intermediate) + endC)
print(cyan + "classRasterSubSampling() : " + endC + "overwrite : " + str(overwrite) + endC)
# Constantes
CODAGE = "uint16"
CODAGE_8B = "uint8"
TEMP = "TempSubSampling_"
MASK_SUF = "_Mask"
SUB_SAMPLE_SUF = "_SubSampled"
CENTROID_SUF = "_Centroids"
TEMP_OUT = "_temp_out"
EXTENSION_TXT = ".txt"
# Contenu de la nouvelle table
text_new_table = ""
# CREATION DES NOMS DE CHEMINS UTILES
name = os.path.splitext(os.path.basename(image_output))[0]
input_classified_image_path = os.path.dirname(classified_image_input) # Ex : D2_Par_Zone/Paysage_01/Corr_2/Resultats/Temp/
temp_sub_sampling_path = input_classified_image_path + os.sep + TEMP + name + os.sep # Dossier contenant les fichiers temporaires de cette brique. Ex : D2_Par_Zone/Paysage_01/Corr_2/Resultats/Temp/Temp_Sub_Sampling/
input_classified_image_complete_name = os.path.basename(classified_image_input) # Ex : Paysage_01_raw.tif
input_classified_image_name = os.path.splitext(input_classified_image_complete_name)[0] # Ex : Paysage_01_raw
input_classified_image_extend = os.path.splitext(input_classified_image_complete_name)[1] # Ex : .tif
image_output_temp = os.path.splitext(image_output)[0] + TEMP_OUT + extension_raster # Ex : D2_Par_Zone/Paysage_01/Corr_2/Resultats/Temp/Temp_Sub_Sampling/Paysage_01_raw_temp.tif
# Création de temp_sub_sampling_path s'il n'existe pas
if not os.path.isdir(os.path.dirname(temp_sub_sampling_path)) :
os.makedirs(os.path.dirname(temp_sub_sampling_path))
print(cyan + "classRasterSubSampling() : " + bold + green + "START ...\n" + endC)
# Lecture du fichier table de proposition
supp_class_list, reaff_class_list, macro_reaff_class_list, sub_sampling_class_list, sub_sampling_number_list = readReallocationTable(table_reallocation, sub_sampling_number) # Fonction de Lib_text
info_table_list = readTextFileBySeparator(table_reallocation, "\n")
# Recherche de la liste des micro classes contenu dans le fichier de classification d'entrée
class_values_list = identifyPixelValues(classified_image_input)
# Supression dans la table des lignes correspondant aux actions "-2"
for ligne_table in info_table_list:
if not "-2" in ligne_table[0]:
text_new_table += str(ligne_table[0]) + "\n"
if debug >= 3:
print("supp_class_list : " + str(supp_class_list))
print("reaff_class_list : " + str(reaff_class_list))
print("macro_reaff_class_list : " + str(macro_reaff_class_list))
print("sub_sampling_class_list : " + str(sub_sampling_class_list))
print("sub_sampling_number_list : " + str(sub_sampling_number_list))
# Dans cettre brique, on ne s'intéresse qu'à la partie sous echantillonage
# Gestion du cas de suppression
if len(supp_class_list) > 0:
print(cyan + "classRasterSubSampling() : " + bold + yellow + "ATTENTION : Les classes ne sont pas supprimees pour le fichier classification format raster." + '\n' + endC)
# Gestion du cas de réaffectation
if len(reaff_class_list) > 0:
print(cyan + "classRasterSubSampling() : " + bold + yellow + "ATTENTION : la brique SpecificSubSampling ne traite pas les reaffectation. A l'issue de cette brique, verifier la table de reallocation et executer la brique de reallocation." + '\n' + endC)
if len(sub_sampling_class_list) > 0 :
if debug >= 3:
print(cyan + "classRasterSubSampling() : " + bold + green + "DEBUT DU SOUS ECHANTILLONAGE DES CLASSES %s " %(sub_sampling_class_list) + endC)
# Parcours des classes à sous échantilloner
processing_pass_first = False
for idx_class in range(len(sub_sampling_class_list)) :
# INITIALISATION DU TRAITEMENT DE LA CLASSE
# Classe à sous échantilloner. Ex : 21008
class_to_sub_sample = sub_sampling_class_list[idx_class]
if idx_class == 0 or not processing_pass_first :
# Image à reclassifier : classified_image_input au premier tour
image_to_sub_sample = classified_image_input
else :
# Image à reclassifier : la sortie de la boucle précédente ensuite
image_to_sub_sample = image_output
# determiner le label disponible de la classe
base_subclass_label = int(class_to_sub_sample/100)*100
subclass_label = base_subclass_label
for class_value in class_values_list:
if (class_value > subclass_label) and (class_value < base_subclass_label + 100) :
subclass_label = class_value
subclass_label += 1
# subclass_label = int(class_to_sub_sample/100)*100 + 20 + class_to_sub_sample%20 * 5
# Label de départ des sous classes. Formule proposée : 3 premiers chiffres de class_to_sub_sample puis ajout de 20 + 5 * class_to_sub_sample modulo 20. Ex : 21000 -> 21020, 21001-> 21025, 21002-> 21030 etc...
# Part du principe qu'il y a moins de 20 micro classes et que chacune est sous échantillonnée au maximum en 5 sous parties. Si ce n'est pas le cas : A ADAPTER
number_of_sub_samples = sub_sampling_number_list[idx_class] # Nombre de sous classes demandées pour le sous échantillonage de class_to_sub_sample. Ex : 4
class_mask_raster = temp_sub_sampling_path + input_classified_image_name + "_" + str(class_to_sub_sample) + MASK_SUF + input_classified_image_extend # Ex : D2_Par_Zone/Paysage_01/Corr_2/Resultats/Temp/Temp_Sub_Sampling/Paysage_01_raw_21008_Mask.tif
class_subsampled_raster = temp_sub_sampling_path + input_classified_image_name + "_" + str(class_to_sub_sample) + SUB_SAMPLE_SUF + input_classified_image_extend # Ex : D2_Par_Zone/Paysage_01/Corr_2/Resultats/Temp/Temp_Sub_Sampling/Paysage_01_raw_21008_SubSampled.tif
centroid_file = temp_sub_sampling_path + input_classified_image_name + "_" + str(class_to_sub_sample) + CENTROID_SUF + EXTENSION_TXT # Ex : D2_Par_Zone/Paysage_01/Corr_2/Resultats/Temp/Temp_Sub_Sampling/Paysage_01_raw_21008_Centroid.txt
if debug >= 5:
print(cyan + "classRasterSubSampling() : " + endC + "class_to_sub_sample :" , class_to_sub_sample)
print(cyan + "classRasterSubSampling() : " + endC + "subclass_label :" , subclass_label)
print(cyan + "classRasterSubSampling() : " + endC + "number_of_sub_samples :" , number_of_sub_samples)
print(cyan + "classRasterSubSampling() : " + endC + "class_mask_raster :" , class_mask_raster)
print(cyan + "classRasterSubSampling() : " + endC + "class_subsampled_raster :" , class_subsampled_raster)
print(cyan + "classRasterSubSampling() : " + endC + "centroid_file :" , centroid_file)
if debug >= 3:
print(cyan + "classRasterSubSampling() : " + bold + green + "CLASSE %s/%s : SOUS ECHANTILLONAGE DE %s EN %s CLASSES " %(idx_class+1, len(sub_sampling_class_list), class_to_sub_sample, number_of_sub_samples) + endC)
# ETAPE 1/5 : EXTRACTION DU MASQUE BINAIRE DES PIXELS CORRESPONDANT A LA CLASSE
expression_masque = "\"im1b1 == %s? 1 : 0\"" %(class_to_sub_sample)
command = "otbcli_BandMath -il %s -out %s %s -exp %s" %(classified_image_input, class_mask_raster, CODAGE_8B, expression_masque)
if debug >=2:
print("\n" + cyan + "classRasterSubSampling() : " + bold + green + "CLASSE %s/%s - ETAPE 1/5 : Debut de l extraction du masque binaire de la classe %s" %(idx_class+1, len(sub_sampling_class_list),class_to_sub_sample) + endC)
print(command)
os.system(command)
if debug >=2:
print(cyan + "classRasterSubSampling() : " + bold + green + "CLASSE %s/%s - ETAPE 1/5 : Fin de l extraction du masque binaire de la classe %s, disponible ici : %s" %(idx_class+1, len(sub_sampling_class_list),class_to_sub_sample, class_mask_raster) + endC)
# TEST POUR SAVOIR SI ON EST EN CAPACITE D'EFFECTUER LE KMEANS
number_of_actives_pixels = countPixelsOfValue(class_mask_raster, 1) # Comptage du nombre de pixels disponibles pour effectuer le kmeans
if number_of_actives_pixels > (number_of_sub_samples * number_of_actives_pixels_threshold) : # Cas où il y a plus de pixels disponibles pour effectuer le kmeans que le seuil
# ETAPE 2/5 : CLASSIFICATION NON SUPERVISEE DES PIXELS CORRESPONDANT A LA CLASSE
if debug >= 3:
print("\n" + cyan + "classRasterSubSampling() : " + bold + green + "CLASSE %s/%s - ETAPE 2/5 : Il y a assez de pixels pour faire le sous echantillonage : %s sur %s requis au minimum " %(idx_class+1, len(sub_sampling_class_list), number_of_actives_pixels, int(number_of_sub_samples) * number_of_actives_pixels_threshold) + endC)
if debug >=2:
print("\n" + cyan + "classRasterSubSampling() : " + bold + green + "CLASSE %s/%s - ETAPE 2/5 : Debut du sous echantillonage par classification non supervisee en %s classes " %(idx_class+1, len(sub_sampling_class_list), number_of_sub_samples) + endC)
# appel du kmeans
input_mask_list = []
input_mask_list.append(class_mask_raster)
output_masked_image_list = []
output_masked_image_list.append(class_subsampled_raster)
output_centroids_files_list = []
output_centroids_files_list.append(centroid_file)
macroclass_sampling_list = []
macroclass_sampling_list.append(number_of_sub_samples)
macroclass_labels_list = []
macroclass_labels_list.append(subclass_label)
applyKmeansMasks(satellite_image_input, input_mask_list, "", "", output_masked_image_list, output_centroids_files_list, macroclass_sampling_list, macroclass_labels_list, no_data_value, path_time_log, 200, 1, -1, 0.0, rand_otb, ram_otb, number_of_actives_pixels_threshold, extension_raster, save_results_intermediate, overwrite)
if debug >=2:
print(cyan + "classRasterSubSampling() : " + bold + green + "CLASSE %s/%s - ETAPE 2/5 : Fin du sous echantillonage par classification non supervisee en %s classes, disponible ici %s : " %(idx_class+1, len(sub_sampling_class_list), number_of_sub_samples, class_subsampled_raster) + endC)
# ETAPE 3/5 : INTEGRATION DES NOUVELLES SOUS CLASSES DANS LA TABLE DE REALLOCATION
# Ouveture du fichier table de proposition pour re-ecriture
for i in range(number_of_sub_samples):
class_values_list.append(subclass_label + i)
text_new_table += str(subclass_label + i) + ";" + str(subclass_label + i) + "; METTRE A JOUR MANUELLEMENT (origine : " + str(class_to_sub_sample) + ")" + "\n"
# ETAPE 4/5 : APPLICATION DU SOUS ECHANTILLONAGE AU RESULTAT DE CLASSIFICATION
expression_application_sous_echantillonage = "\"im1b1 == %s? im2b1 : im1b1\"" %(class_to_sub_sample)
command = "otbcli_BandMath -il %s %s -out %s %s -exp %s" %(image_to_sub_sample, class_subsampled_raster, image_output_temp, CODAGE, expression_application_sous_echantillonage)
if debug >=2:
print("\n" + cyan + "classRasterSubSampling() : " + bold + green + "CLASSE %s/%s - ETAPE 4/5 : Debut de l application du sous echantillonage present dans %s sur %s" %(idx_class+1, len(sub_sampling_class_list), class_subsampled_raster, classified_image_input) + endC)
print(command)
os.system(command)
if debug >=2:
print(cyan + "classRasterSubSampling() : " + bold + green + "CLASSE %s/%s - ETAPE 4/5 : Fin de l application du sous echantillonage present dans %s sur %s, sortie disponible ici : %s" %(idx_class+1, len(sub_sampling_class_list), class_subsampled_raster, classified_image_input, image_output_temp) + endC)
# ETAPE 5/5 : GESTION DES RENOMMAGES ET SUPPRESSIONS
if debug >=2:
print("\n" + cyan + "classRasterSubSampling() : " + bold + green + "CLASSE %s/%s - ETAPE 5/5 : Debut du renommage et suppression des dossiers intermediaires" %(idx_class+1, len(sub_sampling_class_list)) + endC)
if debug >=3 :
print("\n" + green + "classified image input: %s" %(classified_image_input) + endC)
print("\n" + green + "image to sub sample: %s" %(image_to_sub_sample) + endC)
print("\n" + green + "image temp : %s" %(image_output_temp) + endC)
print("\n" + green + "image output : %s" %(image_output) + endC)
# Si l'image d'entrée et l'image de sorte sont le même fichier on efface le fichier d'entrée pour le re-creer avec le fichier re-travaillé
if image_output == classified_image_input and os.path.isfile(classified_image_input) :
removeFile(classified_image_input)
os.rename(image_output_temp,image_output)
processing_pass_first = True
# SUPPRESSION DES FICHIERS TEMPORAIRES
if not save_results_intermediate :
if os.path.isfile(class_mask_raster) :
removeFile(class_mask_raster)
if os.path.isfile(class_subsampled_raster) :
removeFile(class_subsampled_raster)
if os.path.isfile(centroid_file) :
removeFile(centroid_file)
if debug >=2:
print(cyan + "classRasterSubSampling() : " + bold + green + "CLASSE %s/%s - ETAPE 5/5 : Fin du renommage et suppression des dossiers intermediaires" %(idx_class+1, len(sub_sampling_class_list)) + endC)
else: # Cas où il n'y a pas assez de pixels pour effectuer le kmeans
if debug >=2:
print("\n" + cyan + "classRasterSubSampling() : " + bold + yellow + "CLASSE %s/%s - ETAPE 2/5 : Nombre insuffisant de pixels disponibles pour appliquer le kmeans : %s sur %s requis au minimum " %(idx_class+1, len(sub_sampling_class_list), number_of_actives_pixels, int(number_of_sub_samples) * number_of_actives_pixels_threshold) + endC)
print(cyan + "classRasterSubSampling() : " + bold + yellow + "CLASSE %s/%s - ETAPE 2/5 : SOUS ECHANTILLONAGE NON APPLIQUE A LA CLASSE %s" %(idx_class+1, len(sub_sampling_class_list), class_to_sub_sample) + endC + "\n")
# MISE A JOUR DU FICHIER image_to_sub_sample
if idx_class == 0:
processing_pass_first = False
# MISE A JOUR DE LA TABLE DE REALLOCATION
text_new_table += str(class_to_sub_sample) + ";" + str(class_to_sub_sample) + ";CLASSE TROP PETITE POUR SOUS ECHANTILLONAGE" + "\n"
# SUPPRESSION DU MASQUE
if not save_results_intermediate and os.path.isfile(class_mask_raster) :
removeFile(class_mask_raster)
else:
shutil.copy2(classified_image_input, image_output) # Copie du raster d'entree si pas de sous-echantillonnage
# Ecriture de la nouvelle table dans le fichier
writeTextFile(table_reallocation, text_new_table)
# SUPPRESSION DU DOSSIER ET DES FICHIERS TEMPORAIRES
if not save_results_intermediate and os.path.isdir(os.path.dirname(temp_sub_sampling_path)) :
shutil.rmtree(os.path.dirname(temp_sub_sampling_path))
print(cyan + "classRasterSubSampling() : " + bold + green + "END\n" + endC)
# Mise à jour du Log
ending_event = "classRasterSubSampling() : Micro class subsampling on classification image ending : "
timeLine(path_time_log,ending_event)
return
def selectSamples(image_input_list, sample_image_input, vector_output, table_statistics_output, sampler_strategy, select_ratio_floor, ratio_per_class_dico, name_column, no_data_value, path_time_log, rand_seed=0, ram_otb=0, epsg=2154, format_vector='ESRI Shapefile', extension_vector=".shp", save_results_intermediate=False, overwrite=True) :
# Mise à jour du Log
starting_event = "selectSamples() : Select points in raster mask macro input starting : "
timeLine(path_time_log, starting_event)
if debug >= 3:
print(cyan + "selectSamples() : " + endC + "image_input_list : " + str(image_input_list) + endC)
print(cyan + "selectSamples() : " + endC + "sample_image_input : " + str(sample_image_input) + endC)
print(cyan + "selectSamples() : " + endC + "vector_output : " + str(vector_output) + endC)
print(cyan + "selectSamples() : " + endC + "table_statistics_output : " + str(table_statistics_output) + endC)
print(cyan + "selectSamples() : " + endC + "sampler_strategy : " + str(sampler_strategy) + endC)
print(cyan + "selectSamples() : " + endC + "select_ratio_floor : " + str(select_ratio_floor) + endC)
print(cyan + "selectSamples() : " + endC + "ratio_per_class_dico : " + str(ratio_per_class_dico) + endC)
print(cyan + "selectSamples() : " + endC + "name_column : " + str(name_column) + endC)
print(cyan + "selectSamples() : " + endC + "no_data_value : " + str(no_data_value) + endC)
print(cyan + "selectSamples() : " + endC + "path_time_log : " + str(path_time_log) + endC)
print(cyan + "selectSamples() : " + endC + "rand_seed : " + str(rand_seed) + endC)
print(cyan + "selectSamples() : " + endC + "ram_otb : " + str(ram_otb) + endC)
print(cyan + "selectSamples() : " + endC + "epsg : " + str(epsg) + endC)
print(cyan + "selectSamples() : " + endC + "format_vector : " + str(format_vector) + endC)
print(cyan + "selectSamples() : " + endC + "extension_vector : " + str(extension_vector) + endC)
print(cyan + "selectSamples() : " + endC + "save_results_intermediate : " + str(save_results_intermediate) + endC)
print(cyan + "selectSamples() : " + endC + "overwrite : " + str(overwrite) + endC)
# Constantes
EXT_XML = ".xml"
SUFFIX_SAMPLE = "_sample"
SUFFIX_STATISTICS = "_statistics"
SUFFIX_POINTS = "_points"
SUFFIX_VALUE = "_value"
BAND_NAME = "band_"
COLUMN_CLASS = "class"
COLUMN_ORIGINFID = "originfid"
NB_POINTS = "nb_points"
AVERAGE = "average"
STANDARD_DEVIATION = "st_dev"
print(cyan + "selectSamples() : " + bold + green + "DEBUT DE LA SELECTION DE POINTS" + endC)
# Definition variables et chemins
repertory_output = os.path.dirname(vector_output)
filename = os.path.splitext(os.path.basename(vector_output))[0]
sample_points_output = repertory_output + os.sep + filename + SUFFIX_SAMPLE + extension_vector
file_statistic_points = repertory_output + os.sep + filename + SUFFIX_STATISTICS + SUFFIX_POINTS + EXT_XML
if debug >= 3:
print(cyan + "selectSamples() : " + endC + "file_statistic_points : " + str(file_statistic_points) + endC)
# 0. EXISTENCE DU FICHIER DE SORTIE
#----------------------------------
# Si le fichier vecteur points de sortie existe deja et que overwrite n'est pas activé
check = os.path.isfile(vector_output)
if check and not overwrite:
print(bold + yellow + "Samples points already done for file %s and will not be calculated again." %(vector_output) + endC)
else: # Si non ou si la vérification est désactivée : creation du fichier d'échantillons points
# Suppression de l'éventuel fichier existant
if check:
try:
removeVectorFile(vector_output)
except Exception:
pass # Si le fichier ne peut pas être supprimé, on suppose qu'il n'existe pas et on passe à la suite
if os.path.isfile(table_statistics_output) :
try:
removeFile(table_statistics_output)
except Exception:
pass # Si le fichier ne peut pas être supprimé, on suppose qu'il n'existe pas et on passe à la suite
# 1. STATISTIQUE SUR L'IMAGE DES ECHANTILLONS RASTEUR
#----------------------------------------------------
if debug >= 3:
print(cyan + "selectSamples() : " + bold + green + "Start statistique sur l'image des echantillons rasteur..." + endC)
id_micro_list = identifyPixelValues(sample_image_input)
if 0 in id_micro_list :
id_micro_list.remove(0)
min_micro_class_nb_points = -1
min_micro_class_label = 0
infoStructPointSource_dico = {}
writeTextFile(file_statistic_points, '<?xml version="1.0" ?>\n')
appendTextFileCR(file_statistic_points, '<GeneralStatistics>')
appendTextFileCR(file_statistic_points, ' <Statistic name="pointsPerClassRaw">')
if debug >= 2:
print("Nombre de points par micro classe :" + endC)
for id_micro in id_micro_list :
nb_pixels = countPixelsOfValue(sample_image_input, id_micro)
if debug >= 2:
print("MicroClass : " + str(id_micro) + ", nb_points = " + str(nb_pixels))
appendTextFileCR(file_statistic_points, ' <StatisticPoints class="%d" value="%d" />' %(id_micro, nb_pixels))
if min_micro_class_nb_points == -1 or min_micro_class_nb_points > nb_pixels :
min_micro_class_nb_points = nb_pixels
min_micro_class_label = id_micro
infoStructPointSource_dico[id_micro] = StructInfoMicoClass()
infoStructPointSource_dico[id_micro].label_class = id_micro
infoStructPointSource_dico[id_micro].nb_points = nb_pixels
infoStructPointSource_dico[id_micro].info_points_list = []
del nb_pixels
if debug >= 2:
print("MicroClass min points find : " + str(min_micro_class_label) + ", nb_points = " + str(min_micro_class_nb_points))
appendTextFileCR(file_statistic_points, ' </Statistic>')
pending_event = cyan + "selectSamples() : " + bold + green + "End statistique sur l'image des echantillons rasteur. " + endC
if debug >= 3:
print(pending_event)
timeLine(path_time_log,pending_event)
# 2. CHARGEMENT DE L'IMAGE DES ECHANTILLONS
#------------------------------------------
if debug >= 3:
print(cyan + "selectSamples() : " + bold + green + "Start chargement de l'image des echantillons..." + endC)
# Information image
cols, rows, bands = getGeometryImage(sample_image_input)
xmin, xmax, ymin, ymax = getEmpriseImage(sample_image_input)
pixel_width, pixel_height = getPixelWidthXYImage(sample_image_input)
projection_input = getProjectionImage(sample_image_input)
if projection_input == None or projection_input == 0 :
projection_input = epsg
else :
projection_input = int(projection_input)
pixel_width = abs(pixel_width)
pixel_height = abs(pixel_height)
# Lecture des données
raw_data = getRawDataImage(sample_image_input)
if debug >= 3:
print("projection = " + str(projection_input))
print("cols = " + str(cols))
print("rows = " + str(rows))
# Creation d'une structure dico contenent tous les points différents de zéro
progress = 0
pass_prog = False
for y_row in range(rows) :
for x_col in range(cols) :
value_class = raw_data[y_row][x_col]
if value_class != 0 :
infoStructPointSource_dico[value_class].info_points_list.append(x_col + (y_row * cols))
# Barre de progression
if debug >= 4:
if ((float(y_row) / rows) * 100.0 > progress) and not pass_prog :
progress += 1
pass_prog = True
print("Progression => " + str(progress) + "%")
if ((float(y_row) / rows) * 100.0 > progress + 1) :
pass_prog = False
del raw_data
pending_event = cyan + "selectSamples() : " + bold + green + "End chargement de l'image des echantillons. " + endC
if debug >= 3:
print(pending_event)
timeLine(path_time_log,pending_event)
# 3. SELECTION DES POINTS D'ECHANTILLON
#--------------------------------------
if debug >= 3:
print(cyan + "selectSamples() : " + bold + green + "Start selection des points d'echantillon..." + endC)
appendTextFileCR(file_statistic_points, ' <Statistic name="pointsPerClassSelect">')
# Rendre deterministe la fonction aléatoire de random.sample
if rand_seed > 0:
random.seed( rand_seed )
# Pour toute les micro classes
for id_micro in id_micro_list :
# Selon la stategie de selection
nb_points_ratio = 0
while switch(sampler_strategy.lower()):
if case('all'):
# Le mode de selection 'all' est choisi
nb_points_ratio = infoStructPointSource_dico[id_micro].nb_points
infoStructPointSource_dico[id_micro].sample_points_list = range(nb_points_ratio)
break
if case('percent'):
# Le mode de selection 'percent' est choisi
id_macro_class = int(math.floor(id_micro / 100) * 100)
select_ratio_class = ratio_per_class_dico[id_macro_class]
nb_points_ratio = int(infoStructPointSource_dico[id_micro].nb_points * select_ratio_class / 100)
infoStructPointSource_dico[id_micro].sample_points_list = random.sample(range(infoStructPointSource_dico[id_micro].nb_points), nb_points_ratio)
break
if case('mixte'):
# Le mode de selection 'mixte' est choisi
nb_points_ratio = int(infoStructPointSource_dico[id_micro].nb_points * select_ratio_floor / 100)
if id_micro == min_micro_class_label :
# La plus petite micro classe est concervée intégralement
infoStructPointSource_dico[id_micro].sample_points_list = range(infoStructPointSource_dico[id_micro].nb_points)
nb_points_ratio = min_micro_class_nb_points
elif nb_points_ratio <= min_micro_class_nb_points :
# Les micro classes dont le ratio de selection est inferieur au nombre de points de la plus petite classe sont égement conservées intégralement
infoStructPointSource_dico[id_micro].sample_points_list = random.sample(range(infoStructPointSource_dico[id_micro].nb_points), min_micro_class_nb_points)
nb_points_ratio = min_micro_class_nb_points
else :
# Pour toutes les autres micro classes tirage aleatoire d'un nombre de points correspondant au ratio
infoStructPointSource_dico[id_micro].sample_points_list = random.sample(range(infoStructPointSource_dico[id_micro].nb_points), nb_points_ratio)
break
break
if debug >= 2:
print("MicroClass = " + str(id_micro) + ", nb_points_ratio " + str(nb_points_ratio))
appendTextFileCR(file_statistic_points, ' <StatisticPoints class="%d" value="%d" />' %(id_micro, nb_points_ratio))
appendTextFileCR(file_statistic_points, ' </Statistic>')
appendTextFileCR(file_statistic_points, '</GeneralStatistics>')
pending_event = cyan + "selectSamples() : " + bold + green + "End selection des points d'echantillon. " + endC
if debug >= 3:
print(pending_event)
timeLine(path_time_log,pending_event)
# 4. PREPARATION DES POINTS D'ECHANTILLON
#----------------------------------------
if debug >= 3:
print(cyan + "selectSamples() : " + bold + green + "Start preparation des points d'echantillon..." + endC)
# Création du dico de points
points_random_value_dico = {}
index_dico_point = 0
for micro_class in infoStructPointSource_dico :
micro_class_struct = infoStructPointSource_dico[micro_class]
label_class = micro_class_struct.label_class
point_attr_dico = {name_column:int(label_class), COLUMN_CLASS:int(label_class), COLUMN_ORIGINFID:0}
for id_point in micro_class_struct.sample_points_list:
# Recuperer les valeurs des coordonnees des points
coor_x = float(xmin + (int(micro_class_struct.info_points_list[id_point] % cols) * pixel_width)) + (pixel_width / 2.0)
coor_y = float(ymax - (int(micro_class_struct.info_points_list[id_point] / cols) * pixel_height)) - (pixel_height / 2.0)
points_random_value_dico[index_dico_point] = [[coor_x, coor_y], point_attr_dico]
del coor_x
del coor_y
index_dico_point += 1
del point_attr_dico
del infoStructPointSource_dico
pending_event = cyan + "selectSamples() : " + bold + green + "End preparation des points d'echantillon. " + endC
if debug >=3:
print(pending_event)
timeLine(path_time_log,pending_event)
# 5. CREATION DU FICHIER SHAPE DE POINTS D'ECHANTILLON
#-----------------------------------------------------
if debug >= 3:
print(cyan + "selectSamples() : " + bold + green + "Start creation du fichier shape de points d'echantillon..." + endC)
# Définir les attibuts du fichier résultat
attribute_dico = {name_column:ogr.OFTInteger, COLUMN_CLASS:ogr.OFTInteger, COLUMN_ORIGINFID:ogr.OFTInteger}
# Creation du fichier shape
createPointsFromCoordList(attribute_dico, points_random_value_dico, sample_points_output, projection_input, format_vector)
del attribute_dico
del points_random_value_dico
pending_event = cyan + "selectSamples() : " + bold + green + "End creation du fichier shape de points d'echantillon. " + endC
if debug >=3:
print(pending_event)
timeLine(path_time_log,pending_event)
# 6. EXTRACTION DES POINTS D'ECHANTILLONS
#-----------------------------------------
if debug >= 3:
print(cyan + "selectSamples() : " + bold + green + "Start extraction des points d'echantillon dans l'image..." + endC)
# Cas ou l'on a une seule image
if len(image_input_list) == 1:
# Extract sample
image_input = image_input_list[0]
command = "otbcli_SampleExtraction -in %s -vec %s -outfield prefix -outfield.prefix.name %s -out %s -field %s" %(image_input, sample_points_output, BAND_NAME, vector_output, name_column)
if ram_otb > 0:
command += " -ram %d" %(ram_otb)
if debug >= 3:
print(command)
exitCode = os.system(command)
if exitCode != 0:
raise NameError(cyan + "selectSamples() : " + bold + red + "An error occured during otbcli_SampleExtraction command. See error message above." + endC)
# Cas de plusieurs imagettes
else :
# Le repertoire de sortie
repertory_output = os.path.dirname(vector_output)
# Initialisation de la liste pour le multi-threading et la liste de l'ensemble des echantions locaux
thread_list = []
vector_local_output_list = []
# Obtenir l'emprise des images d'entrées pour redecouper le vecteur d'echantillon d'apprentissage pour chaque image
for image_input in image_input_list :
# Definition des fichiers sur emprise local
file_name = os.path.splitext(os.path.basename(image_input))[0]
emprise_local_sample = repertory_output + os.sep + file_name + SUFFIX_SAMPLE + extension_vector
vector_sample_local_output = repertory_output + os.sep + file_name + SUFFIX_VALUE + extension_vector
vector_local_output_list.append(vector_sample_local_output)
# Gestion sans thread...
#SampleLocalExtraction(image_input, sample_points_output, emprise_local_sample, vector_sample_local_output, name_column, BAND_NAME, ram_otb, format_vector, extension_vector, save_results_intermediate)
# Gestion du multi threading
thread = threading.Thread(target=SampleLocalExtraction, args=(image_input, sample_points_output, emprise_local_sample, vector_sample_local_output, name_column, BAND_NAME, ram_otb, format_vector, extension_vector, save_results_intermediate))
thread.start()
thread_list.append(thread)
# Extraction des echantions points des images
try:
for thread in thread_list:
thread.join()
except:
print(cyan + "selectSamples() : " + bold + red + "Erreur lors de l'éextaction des valeurs d'echantion : impossible de demarrer le thread" + endC, file=sys.stderr)
# Fusion des multi vecteurs de points contenant les valeurs des bandes de l'image
fusionVectors(vector_local_output_list, vector_output, format_vector)
# Clean des vecteurs point sample local file
for vector_sample_local_output in vector_local_output_list :
removeVectorFile(vector_sample_local_output)
if debug >= 3:
print(cyan + "selectSamples() : " + bold + green + "End extraction des points d'echantillon dans l'image." + endC)
# 7. CALCUL DES STATISTIQUES SUR LES VALEURS DES POINTS D'ECHANTILLONS SELECTIONNEES
#-----------------------------------------------------------------------------------
if debug >= 3:
print(cyan + "selectSamples() : " + bold + green + "Start calcul des statistiques sur les valeurs des points d'echantillons selectionnees..." + endC)
# Si le calcul des statistiques est demandé presence du fichier stat
if table_statistics_output != "":
# On récupère la liste de données
pending_event = cyan + "selectSamples() : " + bold + green + "Encours calcul des statistiques part1... " + endC
if debug >=4:
print(pending_event)
timeLine(path_time_log,pending_event)
attribute_name_dico = {}
name_field_value_list = []
names_attribut_list = getAttributeNameList(vector_output, format_vector)
if debug >=4:
print("names_attribut_list = " + str(names_attribut_list))
attribute_name_dico[name_column] = ogr.OFTInteger
for name_attribut in names_attribut_list :
if BAND_NAME in name_attribut :
attribute_name_dico[name_attribut] = ogr.OFTReal
name_field_value_list.append(name_attribut)
name_field_value_list.sort()
res_values_dico = getAttributeValues(vector_output, None, None, attribute_name_dico, format_vector)
del attribute_name_dico
# Trie des données par identifiant micro classes
pending_event = cyan + "selectSamples() : " + bold + green + "Encours calcul des statistiques part2... " + endC
if debug >=4:
print(pending_event)
timeLine(path_time_log,pending_event)
data_value_by_micro_class_dico = {}
stat_by_micro_class_dico = {}
# Initilisation du dico complexe
for id_micro in id_micro_list :
data_value_by_micro_class_dico[id_micro] = {}
stat_by_micro_class_dico[id_micro] = {}
for name_field_value in res_values_dico :
if name_field_value != name_column :
data_value_by_micro_class_dico[id_micro][name_field_value] = []
stat_by_micro_class_dico[id_micro][name_field_value] = {}
stat_by_micro_class_dico[id_micro][name_field_value][AVERAGE] = 0.0
stat_by_micro_class_dico[id_micro][name_field_value][STANDARD_DEVIATION] = 0.0
# Trie des valeurs
pending_event = cyan + "selectSamples() : " + bold + green + "Encours calcul des statistiques part3... " + endC
if debug >=4:
print(pending_event)
timeLine(path_time_log,pending_event)
for index in range(len(res_values_dico[name_column])) :
id_micro = res_values_dico[name_column][index]
for name_field_value in name_field_value_list :
data_value_by_micro_class_dico[id_micro][name_field_value].append(res_values_dico[name_field_value][index])
del res_values_dico
# Calcul des statistiques
pending_event = cyan + "selectSamples() : " + bold + green + "Encours calcul des statistiques part4... " + endC
if debug >=4:
print(pending_event)
timeLine(path_time_log,pending_event)
for id_micro in id_micro_list :
for name_field_value in name_field_value_list :
try :
stat_by_micro_class_dico[id_micro][name_field_value][AVERAGE] = average(data_value_by_micro_class_dico[id_micro][name_field_value])
except:
stat_by_micro_class_dico[id_micro][name_field_value][AVERAGE] = 0
try :
stat_by_micro_class_dico[id_micro][name_field_value][STANDARD_DEVIATION] = standardDeviation(data_value_by_micro_class_dico[id_micro][name_field_value])
except:
stat_by_micro_class_dico[id_micro][name_field_value][STANDARD_DEVIATION] = 0
try :
stat_by_micro_class_dico[id_micro][name_field_value][NB_POINTS] = len(data_value_by_micro_class_dico[id_micro][name_field_value])
except:
stat_by_micro_class_dico[id_micro][name_field_value][NB_POINTS] = 0
del data_value_by_micro_class_dico
# Creation du fichier statistique .csv
pending_event = cyan + "selectSamples() : " + bold + green + "Encours calcul des statistiques part5... " + endC
if debug >= 4:
print(pending_event)
timeLine(path_time_log,pending_event)
text_csv = " Micro classes ; Champs couche image ; Nombre de points ; Moyenne ; Ecart type \n"
writeTextFile(table_statistics_output, text_csv)
for id_micro in id_micro_list :
for name_field_value in name_field_value_list :
# Ecriture du fichier
text_csv = " %d " %(id_micro)
text_csv += " ; %s" %(name_field_value)
text_csv += " ; %d" %(stat_by_micro_class_dico[id_micro][name_field_value][NB_POINTS])
text_csv += " ; %f" %(stat_by_micro_class_dico[id_micro][name_field_value][AVERAGE])
text_csv += " ; %f" %(stat_by_micro_class_dico[id_micro][name_field_value][STANDARD_DEVIATION])
appendTextFileCR(table_statistics_output, text_csv)
del name_field_value_list
else :
if debug >=3:
print(cyan + "selectSamples() : " + bold + green + "Pas de calcul des statistiques sur les valeurs des points demander!!!." + endC)
del id_micro_list
pending_event = cyan + "selectSamples() : " + bold + green + "End calcul des statistiques sur les valeurs des points d'echantillons selectionnees. " + endC
if debug >= 3:
print(pending_event)
timeLine(path_time_log,pending_event)
# 8. SUPRESSION DES FICHIERS INTERMEDIAIRES
#------------------------------------------
if not save_results_intermediate:
if os.path.isfile(sample_points_output) :
removeVectorFile(sample_points_output)
print(cyan + "selectSamples() : " + bold + green + "FIN DE LA SELECTION DE POINTS" + endC)
# Mise à jour du Log
ending_event = "selectSamples() : Select points in raster mask macro input ending : "
timeLine(path_time_log,ending_event)
return