def detect_Hough(data):
image = data.copy()
edges = canny(image, sigma=10, low_threshold=60, high_threshold=90)
# Detect circles between 80% and 100% of image semi-diagonal
lx, ly = data.shape
sizex, sizey = lx/2., ly/2.
max_r = numpy.sqrt(sizex**2 + sizey**2)
hough_radii = numpy.linspace(0.5*max_r, 0.9 * max_r, 20)
hough_res = hough_circle(edges, hough_radii)
centers = []
accums = []
radii = []
for radius, h in zip(hough_radii, hough_res):
# For each radius, extract two circles
num_peaks = 2
peaks = peak_local_max(h, num_peaks=num_peaks)
centers.extend(peaks)
accums.extend(h[peaks[:, 0], peaks[:, 1]])
radii.extend([radius] * num_peaks)
# Use the most prominent circle
idx = numpy.argsort(accums)[::-1][:1]
center_x, center_y = centers[idx]
radius = radii[idx]
return center_x, center_y, radius
def generate_sp_and_bd_seg(dataset, INPUT_SIZE):
# Set image paths
image_path = 'images_orig/'
save_path_sp = 'sp_seg_test/'
save_path_bd = 'bd_seg_test/'
img_names = []
# Load image names from txt files
if dataset == 'horsecoarse':
with open('../../lst/' + dataset + '_train.txt', 'r') as f:
img_names.extend([line.strip() for line in f])
with open('../../lst/' + dataset + '_test.txt', 'r') as f:
img_names.extend([line.strip() for line in f])
# Generate and save superpixels and boundaries
for name in img_names:
'''
# Superpixels
img_org = Image.open(image_path+name+'.jpg')
img = np.array(img_org.resize((INPUT_SIZE, INPUT_SIZE)))
labels = segmentation.slic(img, compactness = 40, n_segments = 600)
np.save(save_path_sp+name+"_sp.npy", labels)
'''
# Boundaries
data = np.load("sp_seg/" + name + "_sp.npy")
num_segments = max(np.unique(data))
# Take care of 0th superpixel
bd_dict = {}
labeled = data.copy()
labeled[labeled != 0] = 1
labeled = 1 - labeled
ind = zip(*np.where(
find_boundaries(labeled, mode="inner").astype(np.uint8) == 1))
bd_dict[str(0)] = [str(t) for t in ind]
for i in range(1, num_segments + 1):
labeled = data.copy()
labeled[labeled != i] = 0
ind = zip(*np.where(
find_boundaries(labeled, mode="inner").astype(np.uint8) == 1))
bd_dict[str(i)] = [str(t) for t in ind]
with open(save_path_bd + name + "_bd.json", "w") as f:
json.dump(bd_dict, f)
def parallel_check(coords,data,checkpoint,para,exception,przypadek):
print("parallel_check")
offset=60
print(coords[1][1])
print(coords[3][1])
print(coords[0][0])
print(coords[2][0])
rotated=False
if coords[3][0]+offset>coords[1][0]>coords[3][0]-offset or coords[0][1]+offset>coords[2][1]>coords[0][1]-offset:
data = rotate(data, 5)
checkpoint, contours = background_removal(data.copy())
checkpoint[1] = erosion_loop(checkpoint[1], 5)
coords, przypadek,para,exception = kontury_debug(checkpoint[1])
rotated = True
return coords, przypadek, data,checkpoint,rotated,para,exception
def conservative_smoothing_gray(data, filter_size):
temp = []
indexer = filter_size // 2
new_image = data.copy()
nrow, ncol = data.shape[0:2]
for i in range(nrow):
for j in range(ncol):
for k in range(i - indexer, i + indexer + 1):
for m in range(j - indexer, j + indexer + 1):
if (k > -1) and (k < nrow):
if (m > -1) and (m < ncol):
temp.append(data[k, m])
temp.remove(data[i, j])
max_value = max(temp)
min_value = min(temp)
if data[i, j] > max_value:
new_image[i, j] = max_value
elif data[i, j] < min_value:
new_image[i, j] = min_value
temp = []
return new_image.copy()
def zwroc_pokrojone(data, checkpoint):
data_org = data.copy()
coords, przypadek,para,exception = kontury_debug(checkpoint[1])
coords, przypadek, data, checkpoint, rotated,para,exception = parallel_check(coords, data, checkpoint,para,exception,przypadek)
warp = 0
dst= 0
if przypadek == 0:
warp, dst = transform_if_rectangle(data, coords.copy())
if przypadek == 1:
warp, dst = transform_if_not_rect(data, coords.copy())
tile_img, town_img, road_img, tile_coords, town_coords, road_coords = krojonko(warp, checkpoint)
sheep, forest, clay, mountains, robber, wheat = the_great_tile_finder(tile_img)
domki = the_great_domki_finder(town_img)
print("DEBUG")
print(sheep)
print(forest)
print(clay)
print(mountains)
print(robber)
print(wheat)
print(domki)
print("DEBUG")
# img = data.copy()
# # for i in range(len(img)):
# # for j in range(len(img[0])):
# # img[i][j]=[0,0,0]
# # img=koloruj_piksel(img,sheep,[1,0,0],tile_coords)
# # timestr = time.strftime("%H-%M-%S")
# # io.imsave('debug-' + timestr + '.png', img)
# warp = tf.warp(img, tform3, output_shape=(len(img), len(img[0])))
# sheep, forest, clay, mountains, robber, wheat
# checkpoint.append(warp.copy())
warp = koloruj_tile(warp, sheep, [0, 1, 0], tile_coords)
# checkpoint.append(warp.copy())
warp = koloruj_tile(warp, forest, [0, 1, 1], tile_coords)
# checkpoint.append(warp.copy())
warp = koloruj_tile(warp, clay, [1, 0, 0], tile_coords)
# checkpoint.append(warp.copy())
warp = koloruj_tile(warp, mountains, [0.1, 0.1, 0.1], tile_coords)
# checkpoint.append(warp.copy())
warp = koloruj_tile(warp, robber, [1, 1, 0], tile_coords)
# checkpoint.append(warp.copy())
warp = koloruj_tile(warp, wheat, [1, 0, 1], tile_coords)
# checkpoint.append(warp.copy())
warp = koloruj_town(warp,domki,[1,1,1],town_coords)
warp = reverse_warp(warp, checkpoint, coords, dst)
if rotated:
warp = rotate(warp,-5)
warp = wypelnij(warp,data_org)
# timestr = time.strftime("%H-%M-%S")
# io.imsave('wynik-' + file_string + '-checkpoint-' + str(i) + '-' + timestr + '-warp.png', v)
# checkpoint.append(warp.copy())
ploty = wyswietl(checkpoint, nazwapliku)
ploty[1].plot(coords[:, 1], coords[:, 0], "+r", markersize=15)
ploty[2].plot(town_coords[:, 1], town_coords[:, 0], "+r", markersize=15)
return warp, tile_img, town_img
ploty[2].plot(town_coords[:, 1], town_coords[:, 0], "+r", markersize=15)
return warp, tile_img, town_img
if __name__ == '__main__':
start_time = time.time()
# for file in range(21, 30):
for file in [21,22]:
# for file in range(31, 43):
nazwapliku = str(file) + ".jpg"
print(nazwapliku)
data = io.imread(nazwapliku)
data = img_as_float(data)
checkpoint, contours = background_removal(data.copy())
checkpoint[1] = erosion_loop(checkpoint[1], 5)
warp, tile_img,town_img = zwroc_pokrojone(data, checkpoint)
timestr = time.strftime("%H-%M-%S")
file_string = str(file)
if file < 10:
file_string = '0' + file_string
io.imsave('wynik-' + file_string + '-checkpoint-' + timestr + '-warp.png', warp)
#io.imsave('wynik-' + file_string + '-' + timestr + '-warp.png', warp)
#savefig('wynik-' + file_string + '-' + timestr + '.png')
#for i, v in enumerate(tile_img):
# io.imsave('wynik-' + file_string + '-tile-' + str(i) + '-' + timestr + '-warp.png', v)
# for i, v in enumerate(checkpoint):
from scipy import misc
data = misc.imread('cat.jpeg')
print type(data)
print data.shape
print data.dtype
def rgb2gray(rgb):
r, g, b = rgb[:, :, 0], rgb[:, :, 1], rgb[:, :, 2]
gray = 0.2989 * r + 0.5870 * g + 0.1140 * b
return gray
only_red_value = data.copy()
only_green_value = data.copy()
only_blue_value = data.copy()
only_red_value[:, :, 1:] = [0, 0]
only_green_value[:, :, (0, 2)] = [0, 0]
only_blue_value[:, :, :2] = [0, 0]
gray_scale = data.copy()
gray_scale = rgb2gray(gray_scale)
f, axarr = plt.subplots(3, 2)
print(
"particular point(175,60) at main {} , only red {}, only green {}, only blue {}"
.format(data[175, 60, :], only_red_value[175, 60, :],
only_green_value[175, 60, :], only_blue_value[175, 60, :]))
axarr[0, 0].imshow(data)
axarr[0, 1].imshow(gray_scale, cmap=cm.gray)
def crimmins(data):
new_image = data.copy()
nrow = len(data)
ncol = len(data[0])
# Dark pixel adjustment
# First Step
# N-S
for i in range(1, nrow):
for j in range(ncol):
if data[i - 1, j] >= (data[i, j] + 2):
new_image[i, j] += 1
data = new_image
# E-W
for i in range(nrow):
for j in range(ncol - 1):
if data[i, j + 1] >= (data[i, j] + 2):
new_image[i, j] += 1
data = new_image
# NW-SE
for i in range(1, nrow):
for j in range(1, ncol):
if data[i - 1, j - 1] >= (data[i, j] + 2):
new_image[i, j] += 1
data = new_image
#NE-SW
for i in range(1, nrow):
for j in range(ncol - 1):
if data[i - 1, j + 1] >= (data[i, j] + 2):
new_image[i, j] += 1
data = new_image
# Second Step
# N-S
for i in range(1, nrow - 1):
for j in range(ncol):
if (data[i - 1, j] > data[i, j]) and (data[i, j] <= data[i + 1,
j]):
new_image[i, j] += 1
data = new_image
# E-W
for i in range(nrow):
for j in range(1, ncol - 1):
if (data[i, j + 1] > data[i, j]) and (data[i, j] <= data[i,
j - 1]):
new_image[i, j] += 1
data = new_image
# NW-SE
for i in range(1, nrow - 1):
for j in range(1, ncol - 1):
if (data[i - 1, j - 1] > data[i, j]) and (data[i, j] <=
data[i + 1, j + 1]):
new_image[i, j] += 1
data = new_image
# NE-SW
for i in range(1, nrow - 1):
for j in range(1, ncol - 1):
if (data[i - 1, j + 1] > data[i, j]) and (data[i, j] <=
data[i + 1, j - 1]):
new_image[i, j] += 1
data = new_image
#Third Step
# N-S
for i in range(1, nrow - 1):
for j in range(ncol):
if (data[i + 1, j] > data[i, j]) and (data[i, j] <= data[i - 1,
j]):
new_image[i, j] += 1
data = new_image
# E-W
for i in range(nrow):
for j in range(1, ncol - 1):
if (data[i, j - 1] > data[i, j]) and (data[i, j] <= data[i,
j + 1]):
new_image[i, j] += 1
data = new_image
# NW-SE
for i in range(1, nrow - 1):
for j in range(1, ncol - 1):
if (data[i + 1, j + 1] > data[i, j]) and (data[i, j] <=
data[i - 1, j - 1]):
new_image[i, j] += 1
data = new_image
# NE-SW
for i in range(1, nrow - 1):
for j in range(1, ncol - 1):
if (data[i + 1, j - 1] > data[i, j]) and (data[i, j] <=
data[i - 1, j + 1]):
new_image[i, j] += 1
data = new_image
# Fourth Step
# N-S
for i in range(nrow - 1):
for j in range(ncol):
if (data[i + 1, j] >= (data[i, j] + 2)):
new_image[i, j] += 1
data = new_image
# E-W
for i in range(nrow):
for j in range(1, ncol):
if (data[i, j - 1] >= (data[i, j] + 2)):
new_image[i, j] += 1
data = new_image
# NW-SE
for i in range(nrow - 1):
for j in range(ncol - 1):
if (data[i + 1, j + 1] >= (data[i, j] + 2)):
new_image[i, j] += 1
data = new_image
# NE-SW
for i in range(nrow - 1):
for j in range(1, ncol):
if (data[i + 1, j - 1] >= (data[i, j] + 2)):
new_image[i, j] += 1
data = new_image
# Light pixel adjustment
# First Step
# N-S
for i in range(1, nrow):
for j in range(ncol):
if (data[i - 1, j] <= (data[i, j] - 2)):
new_image[i, j] -= 1
data = new_image
# E-W
for i in range(nrow):
for j in range(ncol - 1):
if (data[i, j + 1] <= (data[i, j] - 2)):
new_image[i, j] -= 1
data = new_image
# NW-SE
for i in range(1, nrow):
for j in range(1, ncol):
if (data[i - 1, j - 1] <= (data[i, j] - 2)):
new_image[i, j] -= 1
data = new_image
# NE-SW
for i in range(1, nrow):
for j in range(ncol - 1):
if (data[i - 1, j + 1] <= (data[i, j] - 2)):
new_image[i, j] -= 1
data = new_image
# Second Step
# N-S
for i in range(1, nrow - 1):
for j in range(ncol):
if (data[i - 1, j] < data[i, j]) and (data[i, j] >= data[i + 1,
j]):
new_image[i, j] -= 1
data = new_image
# E-W
for i in range(nrow):
for j in range(1, ncol - 1):
if (data[i, j + 1] < data[i, j]) and (data[i, j] >= data[i,
j - 1]):
new_image[i, j] -= 1
data = new_image
# NW-SE
for i in range(1, nrow - 1):
for j in range(1, ncol - 1):
if (data[i - 1, j - 1] < data[i, j]) and (data[i, j] >=
data[i + 1, j + 1]):
new_image[i, j] -= 1
data = new_image
# NE-SW
for i in range(1, nrow - 1):
for j in range(1, ncol - 1):
if (data[i - 1, j + 1] < data[i, j]) and (data[i, j] >=
data[i + 1, j - 1]):
new_image[i, j] -= 1
data = new_image
# Third Step
# N-S
for i in range(1, nrow - 1):
for j in range(ncol):
if (data[i + 1, j] < data[i, j]) and (data[i, j] >= data[i - 1,
j]):
new_image[i, j] -= 1
data = new_image
# E-W
for i in range(nrow):
for j in range(1, ncol - 1):
if (data[i, j - 1] < data[i, j]) and (data[i, j] >= data[i,
j + 1]):
new_image[i, j] -= 1
data = new_image
# NW-SE
for i in range(1, nrow - 1):
for j in range(1, ncol - 1):
if (data[i + 1, j + 1] < data[i, j]) and (data[i, j] >=
data[i - 1, j - 1]):
new_image[i, j] -= 1
data = new_image
# NE-SW
for i in range(1, nrow - 1):
for j in range(1, ncol - 1):
if (data[i + 1, j - 1] < data[i, j]) and (data[i, j] >=
data[i - 1, j + 1]):
new_image[i, j] -= 1
data = new_image
# Fourth Step
# N-S
for i in range(nrow - 1):
for j in range(ncol):
if (data[i + 1, j] <= (data[i, j] - 2)):
new_image[i, j] -= 1
data = new_image
# E-W
for i in range(nrow):
for j in range(1, ncol):
if (data[i, j - 1] <= (data[i, j] - 2)):
new_image[i, j] -= 1
data = new_image
# NW-SE
for i in range(nrow - 1):
for j in range(ncol - 1):
if (data[i + 1, j + 1] <= (data[i, j] - 2)):
new_image[i, j] -= 1
data = new_image
# NE-SW
for i in range(nrow - 1):
for j in range(1, ncol):
if (data[i + 1, j - 1] <= (data[i, j] - 2)):
new_image[i, j] -= 1
data = new_image
return new_image.copy()