def find_time_without_value(dict_shift_value, list_mms):
height_needed = []
for fn in list_mms:
neighbours = search_neigh_tiles(fn)
for nfn in neighbours:
if nfn not in dict_shift_value or np.isnan(dict_shift_value[nfn]):
height_needed.append(nfn)
return list(set(height_needed))
def insert_values_based_on_neigbours(dict_shift_value, height_needed):
ndict = dict()
for fn in height_needed:
neighbours = search_neigh_tiles(fn, radius=1)
z = get_distance_without_padding(dict_shift_value, neighbours)
if len(z) > 0:
if not np.isnan(np.median(z)):
ndict[fn] = np.median(z)
for key in ndict.keys():
dict_shift_value[key] = ndict[key]
return dict_shift_value
def gen_updates4mms(list_fns, dict_shift_value, r, res_update, shift):
from lib.neighbour import search_neigh_tiles
# Tile Size in meter
size = int(r / res_update)
# Container of upadte images
dict_update_img = dict()
# Container of upadte points
dict_update_pts = dict()
# Collect updates by bilinear intepolation
for fn in tqdm(list_fns):
# 9 Neighbour tiles
neigh_list = search_neigh_tiles(fn)
data_update, z = generate_updates_mms(dict_shift_value, neigh_list, r,
res_update, shift)
# Save image
if type(data_update) != type(None):
update_img = data_update[:, 2].reshape(size, size)
dict_update_img[fn] = update_img
dict_update_pts[fn] = data_update
return dict_update_img, dict_update_pts
# Remove because for runids
#def update_pointcloud(data_mms, data_update, res_update):
#
# # Create index according to updates resolution
# d_update = rasterize(data_update, res_update, dim = 2)
# d_mms_runid = rasterize(data_mms, res_update, dim = 2)
#
# # Apply change for each cell regarding updates resolution
# data_updated = []
# for key, value in d_mms_runid.iteritems():
# sub_mms = data_mms[value]
# if key in d_update:
# update_value = data_update[d_update[key][0],2]
# sub_mms[:,:3] = sub_mms[:,:3] - [0,0,update_value]
# data_updated.extend(sub_mms) # Must be minus here
# return np.array(data_updated)
def reject_local_change(dict_shift_value, list_mms, thre = 0.5):
# Reject local changes over threshold
tobedeleted = []
for fn in list_mms:
neighbours = search_neigh_tiles(fn, radius = 2)
data = [[nfn, dict_shift_value[nfn] - shift] for nfn in neighbours if nfn in dict_shift_value.keys()]
fns = np.array(column(data, 0))
z = np.array(column(data, 1))
idx = abs(z - np.median(z)) > thre
if (True) in idx:
# print(list(fns[idx]), list(z[idx]))
# print(list(z))
tobedeleted.extend(fns[idx])
tobedeleted = list(set(tobedeleted))
print(len(tobedeleted))
for fn in tobedeleted:
dict_shift_value.pop(fn)
return dict_shift_value, len(tobedeleted)
x, y = [int(k) for k in key.split('+')]
z_mms = data_mms[value, 2]
z_ref = data_ref[d_ref[key], 2][0] + shift
dz = z_mms - z_ref
newIndex = list(np.array(value)[dz < 0.5])
index_list.extend(newIndex)
data_output = data_mms[sorted(index_list)]
data_mms = data_output
_ = calc_difference_mms_ref(data_mms, data_ref, 0.5, r, shift)
neigh_list = search_neigh_tiles(fn)
data_update, z = generate_updates_mms(dict_shift_value, neigh_list, r,
res_update, shift)
d_update = rasterize(data_update, res_update, dim=2)
d_mms = rasterize(data_mms, res_update, dim=2)
data_updated = []
for key, value in d_mms.iteritems():
sub_mms = data_mms[value]
update_value = data_update[d_update[key][0], 2]
data_updated.extend(sub_mms -
[0, 0, update_value]) # Must be minus here
data_updated = np.array(data_updated)
_ = calc_difference_mms_ref(data_updated, data_ref, 0.5, r, shift)