def upload_files():
files = request.files
title = request.args.get('title')
response_list = []
for file_object in files.getlist('user_file[]'):
# file size
file_size = get_file_size(file_object)
if file_size > FILE_SIZE_THRESHOLD:
continue
# file_mimetype
file_mimetype = file_object.mimetype
if file_mimetype not in ACCEPTED_TYPES_LIST:
continue
# file name
file_name = file_object.filename
# file extension
file_extension = file_object.filename.rsplit('.', 1)[-1]
# update db and upload file upload file
try:
upload_to_s3(s3_bucket_object, file_object, s3_bucket_name)
file_schema = S3File(extension=file_extension,
title=title,
name=file_name,
mimeType=file_mimetype,
size=file_size,
webLink=S3_WEBLINK_STRUCTURE.format(
bucket_name=s3_bucket_name,
file_name=file_name))
db.session.add(file_schema)
db.session.commit()
response_object = S3File.query.filter_by(name=file_name).first()
response_object = response_object.__dict__
del response_object['_sa_instance_state']
response_list.append(response_object)
except Exception:
pass
final_response_object = {'file_objects': response_list}
return jsonify(final_response_object)
def get_geojson_carto_gs(self,
filepath: tuple = (),
opts: dict = {}) -> dict:
# [Step - 1] Get image + check color sampling
img_path = os.path.join(BASE_DIR, filepath[0])
img_tiff_path = os.path.join(BASE_DIR, filepath[1])
img_extension = os.path.splitext(img_path)[1]
img_name = ntpath.basename(img_path).replace(img_extension, '')
img_base_path = img_path.replace(ntpath.basename(img_path), '')
color_preset = self.data['color_presets'][self.options['color_preset']]
logger.info('Color Preset (Carto Grayscale): ',
{'color_preset': color_preset})
do_contour_normalization = bool(
color_preset['building']['normalize_contours']
) if 'normalize_contours' in color_preset['building'] else False
image = cv2.imread(img_path, 1)
fc_bgr_building_gray = color_preset['building']['fill']['gray']
fc_hsv_building_gray = bgr_color_to_hsv(fc_bgr_building_gray)
if color_preset['building']['border']['type'] == 'relative':
fc_hsv_building_gray_darker = self.transform_relative_color(
fc_hsv_building_gray,
color_preset['building']['border']['value']['gray'])
else:
fc_hsv_building_gray_darker = self.transform_color_string_to_float(
color_preset['building']['border']['value']['gray'])
logger.debug(
self.logger_base_text + 'Color Info', {
'fill_color_bgr': {
'gray': fc_bgr_building_gray
},
'fill_color_hsv': {
'gray': fc_hsv_building_gray
},
'border_color_hsv': {
'gray': fc_hsv_building_gray_darker
}
})
# [Step-2] Do masking on HSV Image
img_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
hsv = cv2.cvtColor(img_rgb, cv2.COLOR_RGB2HSV)
mask_gray = cv2.inRange(hsv, fc_hsv_building_gray,
fc_hsv_building_gray_darker)
final = cv2.bitwise_or(image, image, mask=mask_gray)
# [Step-3] Find Contours
json_contour_filepath = self.data['file']['json_contour'].replace(
'<result_path>', self.data['path']['result']).replace(
'<img_name>', img_name).replace('<preset>', 'carto-gs')
json_contour_debug_filepath = self.data['file'][
'json_contour_debug'].replace('<result_path>',
self.data['path']['result']).replace(
'<img_name>', img_name).replace(
'<preset>', 'carto-gs')
geojson_filepath = self.data['file']['geojson'].replace(
'<result_path>', self.data['path']['result']).replace(
'<img_name>', img_name).replace('<preset>', 'carto-gs')
final_gray = cv2.cvtColor(final, cv2.COLOR_BGR2GRAY)
final_blurred = cv2.GaussianBlur(final_gray, (3, 3), 0)
ret, final_thresh = cv2.threshold(final_blurred, 127, 255, 0)
contours, hierarchy = cv2.findContours(final_thresh, cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
# contour normalization
if do_contour_normalization:
contours = self.normalize_contours(contours)
ctr_json_str = json.dumps(
{
'contours': contours,
'hierarchy': hierarchy
},
default=json_np_default_parser)
ctr_json = json.loads(ctr_json_str)
ctr_points = []
for cidx in range(len(ctr_json['contours'])):
ctr_points.append(
list(map(lambda x: x[0], ctr_json['contours'][cidx])))
# [Step - 4] Find Contours Geographic Coordinates
geotiff_image = img_path.replace(img_extension, '.tif')
translate_coords = GeoTiffProcessor.get_multi_polygon_axis_point_coordinates(
geotiff_image, ctr_points, {'debug': False})
final_coords = []
geo_features = []
for poly in translate_coords['coords']:
poly_coords = []
poly_geo_coords = []
for cr in poly:
poly_coords.append({
'x': cr['x'],
'y': cr['y'],
'latitude': cr['lat'],
'longitude': cr['long']
})
poly_geo_coords.append((cr['long'], cr['lat']))
# add final closing point
poly_geo_coords.append((poly[0]['long'], poly[0]['lat']))
final_coords.append(poly_coords)
geo_feature = Feature(
geometry=Polygon([poly_geo_coords], precision=15))
geo_features.append(geo_feature)
geo_feature_collection = FeatureCollection(geo_features)
geo_feature_collection_dump = geojson_dumps(geo_feature_collection,
sort_keys=True)
with open(json_contour_filepath, 'w') as outfile:
json.dump(final_coords, outfile)
with open(geojson_filepath, 'w') as outfile:
outfile.write(geo_feature_collection_dump)
# [Step-5] Draw contours to original image clone
final_wctrs = copy(image)
for c in contours:
cv2.drawContours(final_wctrs, [c], 0,
color_preset['building']['contour'], 2)
# Build result
polygon_len = len(ctr_points)
r = {
'file_path': geojson_filepath,
'file_size':
str(get_file_size(geojson_filepath, SIZE_UNIT.KB)) + ' KB',
'polygon_total': polygon_len
}
if 'return_polygon_data' in opts and bool(opts['return_polygon_data']):
r['geojson'] = json.loads(geo_feature_collection_dump)
if self.options['save_result']:
result_ftemplate = self.data['path'][
'result'] + img_name + '-carto-gs-<fnm>' + img_extension
self.write_image_results(
result_ftemplate, '<fnm>',
[('step-1-2-hsv-building-gray', fc_hsv_building_gray),
('step-2-image-bgr', image), ('step-3-image-rgb', img_rgb),
('step-4-0-hsv', hsv), ('step-4-1-hsv-mask-gray', mask_gray),
('step-5-final', final), ('step-6-image-gray', final_gray),
('step-7-final-blurred', final_blurred),
('step-8-final-thresh', final_thresh),
('step-9-image-final-with-contours', final_wctrs)])
if self.options['show_result']:
show_image_results([
("Step - 1-1 (HSV Gray Color)",
np.uint8([[fc_hsv_building_gray]])),
("Step - 2 (Image - BGR)", image),
("Step - 3 ( Image - RGB)", img_rgb),
("Step - 4-0 (HSV)", hsv),
("Step - 4-1 (HSV - Gray)", mask_gray),
("Step - 5 (Final)", final),
("Step - 6 (Final - Gray)", final_gray),
("Step - 7 (Final - Gray Blurred)", final_blurred),
("Step - 8 (Final - Gray Thresh)", final_thresh),
("Step - 9 (Final - with contours)", final_wctrs)
])
# [Step - ending] Clean - up
del contours, hierarchy, image, img_rgb, hsv, final, final_gray, final_wctrs, final_blurred, final_thresh, mask_gray, fc_hsv_building_gray
return r
else:
# [Step - ending] Clean - up
del contours, hierarchy, image, img_rgb, hsv, final, final_gray, final_wctrs, final_blurred, final_thresh, mask_gray, fc_hsv_building_gray
return r
def get_geojson_osm(self, filepath: tuple = (), opts: dict = {}) -> dict:
# [Step - 1] Get image + check color sampling
img_path = os.path.join(BASE_DIR, filepath[0])
img_tiff_path = os.path.join(BASE_DIR, filepath[1])
img_extension = os.path.splitext(img_path)[1]
img_name = ntpath.basename(img_path).replace(img_extension, '')
img_base_path = img_path.replace(ntpath.basename(img_path), '')
color_preset = self.data['color_presets'][self.options['color_preset']]
logger.info('Color Preset (OSM): ', {'color_preset': color_preset})
do_contour_normalization = bool(
color_preset['building']['normalize_contours']
) if 'normalize_contours' in color_preset['building'] else False
image_origin = cv2.imread(img_path, 1)
if 'sharp_image' in color_preset['building']:
sharp_img = self.unsharp_mask(
image_origin, **color_preset['building']['sharp_image'])
image_origin = copy(image_origin)
image_new_contrast = []
if 'adjust_contrast' in color_preset['building']:
image = cv2.convertScaleAbs(
image_origin,
alpha=color_preset['building']['adjust_contrast']['alpha'],
beta=color_preset['building']['adjust_contrast']['beta'])
image_new_contrast = [
cv2.convertScaleAbs(image_origin, alpha=1.0, beta=-10),
cv2.convertScaleAbs(image_origin, alpha=1.0, beta=-20),
cv2.convertScaleAbs(image_origin, alpha=1.0, beta=-30),
cv2.convertScaleAbs(image_origin, alpha=1.0, beta=-50),
cv2.convertScaleAbs(image_origin, alpha=1.0, beta=-60)
]
else:
image = copy(image_origin)
light_brown = np.uint8([[color_preset['building']['fill']]])
# Enhance image (ref: https://chrisalbon.com/machine_learning/preprocessing_images/enhance_contrast_of_greyscale_image/)
# image = cv2.imread('images/plane_256x256.jpg', cv2.IMREAD_GRAYSCALE)
# image_enhanced = cv2.equalizeHist(image)
# Convert BGR to HSV for masking
color_codes = []
hsv_fill_color = cv2.cvtColor(light_brown, cv2.COLOR_BGR2HSV)
# hsv_fill_color = cv2.cvtColor(light_brown, color_preset['building']['masking_color_mode'])
# for index in hsv_fill_color:
# color_codes = index[0]
color_codes = hsv_fill_color[0][0]
# [Step - 2] Do masking on HSV Image
img_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
# img_rgb =copy(image)
hsv = cv2.cvtColor(img_rgb, cv2.COLOR_RGB2HSV)
# hsv = cv2.cvtColor(img_rgb, color_preset['building']['masking_color_mode'])
fill_color = (float(color_codes[0]), float(color_codes[1]),
float(color_codes[2]))
find_border_color = []
if color_preset['building']['border']['type'] == 'relative':
temp = []
for idx, bbv in enumerate(
color_preset['building']['border']['value'], 0):
if bbv[0] == '+':
temp.append(float(color_codes[idx]) + float(bbv[1:]))
elif bbv[0] == '-':
temp.append(float(color_codes[idx]) - float(bbv[1:]))
else:
temp.append(float(bbv))
border_color = tuple(temp)
# border_color = (float(color_codes[0]) + color_preset['building']['border']['value'][0], float(color_codes[1]) + color_preset['building']['border']['value'][1], float(color_codes[2]) + color_preset['building']['border']['value'][2])
else:
find_border_color = cv2.cvtColor(
np.uint8([[color_preset['building']['border']['value']]]),
cv2.COLOR_BGR2HSV)
# find_border_color = cv2.cvtColor(np.uint8([[color_preset['building']['border']['value']]]), color_preset['building']['masking_color_mode'])
border_color = (float(find_border_color[0][0][0]),
float(find_border_color[0][0][1]),
float(find_border_color[0][0][2]))
logger.debug(
self.logger_base_text + 'Color Info', {
'fill_color': fill_color,
'border_color': border_color,
'float_border_color': find_border_color,
'hsv_fill_color_codes': color_codes,
'hsv_fill_color': hsv_fill_color
})
mask = cv2.inRange(hsv, fill_color, border_color)
final = cv2.bitwise_and(image, image, mask=mask)
# self.data['path']['result']
# self.data['file']['json_contour']
json_contour_filepath = self.data['file']['json_contour'].replace(
'<result_path>', self.data['path']['result']).replace(
'<img_name>', img_name).replace('<preset>', 'osm')
json_contour_debug_filepath = self.data['file'][
'json_contour_debug'].replace('<result_path>',
self.data['path']['result']).replace(
'<img_name>', img_name).replace(
'<preset>', 'osm')
geojson_filepath = self.data['file']['geojson'].replace(
'<result_path>', self.data['path']['result']).replace(
'<img_name>', img_name).replace('<preset>', 'osm')
final_gray = cv2.cvtColor(final, cv2.COLOR_BGR2GRAY)
final_blurred = cv2.GaussianBlur(final_gray, (5, 5), 0)
ret, final_thresh = cv2.threshold(final_blurred, 127, 255, 0)
contours, hierarchy = cv2.findContours(final_thresh, cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
# contour normalization
if do_contour_normalization:
contours = self.normalize_contours(contours)
ctr_json_str = json.dumps(
{
'contours': contours,
'hierarchy': hierarchy
},
default=json_np_default_parser)
ctr_json = json.loads(ctr_json_str)
ctr_points = []
for cidx in range(len(ctr_json['contours'])):
ctr_points.append(
list(map(lambda x: x[0], ctr_json['contours'][cidx])))
# [Step - 4] Find Contours Geographic Coordinates
geotiff_image = img_tiff_path
translate_coords = GeoTiffProcessor.get_multi_polygon_axis_point_coordinates(
geotiff_image, ctr_points, {'debug': False})
final_coords = []
geo_features = []
for poly in translate_coords['coords']:
poly_coords = []
poly_geo_coords = []
for cr in poly:
poly_coords.append({
'x': cr['x'],
'y': cr['y'],
'latitude': cr['lat'],
'longitude': cr['long']
})
poly_geo_coords.append((cr['long'], cr['lat']))
# add final closing point
poly_geo_coords.append((poly[0]['long'], poly[0]['lat']))
final_coords.append(poly_coords)
geo_feature = Feature(
geometry=Polygon([poly_geo_coords], precision=15))
geo_features.append(geo_feature)
geo_feature_collection = FeatureCollection(geo_features)
geo_feature_collection_dump = geojson_dumps(geo_feature_collection,
sort_keys=True)
with open(json_contour_filepath, 'w') as outfile:
json.dump(final_coords, outfile)
with open(geojson_filepath, 'w') as outfile:
outfile.write(geo_feature_collection_dump)
# [Step - 5] Draw contours to original image clone
final_wctrs = copy(
image
) # final_wctrs = copy(image_origin)# final_wctrs = copy(final)
for c in contours:
cv2.drawContours(final_wctrs, [c], 0,
color_preset['building']['contour'], 2)
# Build result
polygon_len = len(ctr_points)
r = {
'file_path': geojson_filepath,
'file_size':
str(get_file_size(geojson_filepath, SIZE_UNIT.KB)) + ' KB',
'polygon_total': polygon_len
}
if 'return_polygon_data' in opts and bool(opts['return_polygon_data']):
r['geojson'] = json.loads(geo_feature_collection_dump)
if self.options['save_result']:
result_ftemplate = self.data['path'][
'result'] + img_name + '-<fnm>' + img_extension
if 'sharp_image' in color_preset['building']:
cv2.imwrite(
result_ftemplate.replace('<fnm>', 'step-0-sharpen-1'),
sharp_img)
if 'adjust_contrast' in color_preset['building']:
cv2.imwrite(
result_ftemplate.replace('<fnm>', 'step-0-contrast-1'),
image_new_contrast[0])
cv2.imwrite(
result_ftemplate.replace('<fnm>', 'step-0-contrast-2'),
image_new_contrast[1])
cv2.imwrite(
result_ftemplate.replace('<fnm>', 'step-0-contrast-3'),
image_new_contrast[2])
cv2.imwrite(
result_ftemplate.replace('<fnm>', 'step-0-contrast-4'),
image_new_contrast[3])
cv2.imwrite(
result_ftemplate.replace('<fnm>', 'step-0-contrast-5'),
image_new_contrast[4])
cv2.imwrite(
result_ftemplate.replace('<fnm>', 'step-1-hsv-light-color'),
hsv_fill_color)
cv2.imwrite(result_ftemplate.replace('<fnm>', 'step-2-image-bgr'),
image)
cv2.imwrite(result_ftemplate.replace('<fnm>', 'step-3-image-rgb'),
img_rgb)
cv2.imwrite(result_ftemplate.replace('<fnm>', 'step-4-hsv'), hsv)
cv2.imwrite(result_ftemplate.replace('<fnm>', 'step-5-final'),
final)
cv2.imwrite(result_ftemplate.replace('<fnm>', 'step-6-image-gray'),
final_gray)
cv2.imwrite(
result_ftemplate.replace('<fnm>', 'step-7-final-blurred'),
final_blurred)
cv2.imwrite(
result_ftemplate.replace('<fnm>', 'step-8-final-thresh'),
final_thresh)
cv2.imwrite(
result_ftemplate.replace('<fnm>',
'step-9-image-final-with-contours'),
final_wctrs)
if self.options['show_result']:
cv2.imshow("Step - 1 (HSV Light Color)", hsv_fill_color)
cv2.imshow("Step - 2 (Image - BGR)", image)
cv2.imshow("Step - 3 ( Image - RGB)", img_rgb)
cv2.imshow("Step - 4 (HSV)", hsv)
cv2.imshow("Step - 5 (Final)", final)
cv2.imshow("Step - 6 (Final - Gray)", final_gray)
cv2.imshow("Step - 7 (Final - Gray Blurred)", final_blurred)
cv2.imshow("Step - 8 (Final - Gray Thresh)", final_thresh)
cv2.imshow("Step - 9 (Final - with contours)", final_wctrs)
# cv2.imshow("Step - 10 (Final - with shape contours)", final_shape_ctrs)
cv2.waitKey(0)
cv2.destroyAllWindows()
# [Step - ending] Clean - up
del contours, hierarchy, image, hsv_fill_color, img_rgb, hsv, final, final_gray, final_wctrs, final_blurred, final_thresh, ctr_json, ctr_json_str, final_coords, geo_features, ctr_points
return r
else:
# [Step - ending] Clean - up
del contours, hierarchy, image, hsv_fill_color, img_rgb, hsv, final, final_gray, final_wctrs, final_blurred, final_thresh, ctr_json, ctr_json_str, final_coords, geo_features, ctr_points
return r