def main():
print(len(all_info), 'vases')
link_dict = dict()
for img_id, img_info in tqdm(all_info.items()):
# input(img_info) # if you want to see one line
if 'fragment' in img_info['Title'].lower() or \
'fragment' in img_info['description'].lower() or \
'Fragments' in img_info['categories']:
link_dict[img_id] = img_info['src']
try:
img = dip.imread(id_to_img_name(img_id))
except:
continue
print(img.shape)
new_img = dip.resize(img, dsize=outsize, interpolation=INTER_AREA)
dip.im_write(new_img, id_to_out_name(img_id))
elif 'terra' in img_info['Title'].lower() or \
'Medium' not in img_info or \
('Medium' in img_info and 'terra' in img_info['Medium'].lower()) or \
'terra' in img_info['description'].lower() or \
'Terracotta' in img_info['categories']:
pass
# if outsize:
# try:
# img = dip.imread(id_to_img_name(img_id))
# except:
# continue
# print(img.shape)
# new_img = dip.resize(img, dsize=outsize, interpolation=INTER_AREA)
# dip.im_write(new_img, id_to_out_name(img_id))
# else:
# shutil.copyfile(id_to_img_name(img_id), id_to_out_name(img_id))
with open('data/raw/frag_links.pkl', 'wb') as f:
pickle.dump(link_dict, f)
def main(outsize=default_outsize):
print(len(all_info), 'vases')
for img_id, img_info in all_info.items():
# input(img_info) # if you want to see one line
if 'fragment' in img_info['Title'].lower() or \
'fragment' in img_info['description'].lower() or \
'Fragments' in img_info['categories']:
continue
# just get Terracotta
elif 'terra' in img_info['Title'].lower() or \
'Medium' not in img_info or \
('Medium' in img_info and 'terra' in img_info['Medium'].lower()) or \
'terra' in img_info['description'].lower() or \
'Terracotta' in img_info['categories']:
# just get everything that's not broken
# else:
if outsize:
try:
img = dip.imread(id_to_img_name(img_id))
except:
continue
print(img.shape)
new_img = dip.resize(img,
dsize=outsize,
interpolation=INTER_AREA)
dip.im_write(new_img, id_to_out_name(img_id))
else:
shutil.copyfile(id_to_img_name(img_id), id_to_out_name(img_id))
def main_pix2pix():
def out_pix2pix_name():
global _pix2pix_counter
outname = f'{_pix2pix_dir}/{_pix2pix_counter}.jpg'
_pix2pix_counter += 1
return outname
for f_img in glob.glob(dir_in + '/*'):
try:
img = dip.imread(f_img)
except:
continue
img_out = dip.resize(img, _pix2pix_outsize, interpolation=INTER_AREA)
print(f_img, img.shape)
if len(img.shape) == 3:
gray = np.mean(img, axis=-1)
else:
# gray = img
continue
grad = dip.transforms.edge_detect(gray)
mask, m_min, m_max, n_min, n_max = space_fill(grad)
for n in range(n_fragments):
# frag_size = max(img.shape[0], img.shape[1]) // 4
frag_size = img.shape[0]//4, img.shape[1]//4
frag, _ = fragment(img, m_min, m_max, n_min, n_max, frag_size=frag_size)
if frag is None:
break
else:
frag = dip.resize(frag, _pix2pix_outsize, interpolation=INTER_AREA)
both = np.concatenate([img_out, frag], axis=1)
# plt.imshow(both)
# plt.show()
dip.im_write(both, out_pix2pix_name())
def main_biggan():
for f_img in glob.glob(dir_in + '/*'):
img = dip.imread(f_img)
img_id = int(os.path.split(f_img)[-1].split('.')[0])
print(f_img, img.shape)
if len(img.shape) == 3:
gray = np.mean(img, axis=-1)
else:
# gray = img
continue
grad = dip.transforms.edge_detect(gray)
mask, m_min, m_max, n_min, n_max = space_fill(grad)
grad = mark_image_box(grad, m_min, m_max, n_min, n_max)
frag_failed = False
for n in range(n_fragments):
frag, _ = fragment(img, m_min, m_max, n_min, n_max)
if frag is None:
frag_failed = True
break
dip.im_write(frag, out_frag(img_id, n))
if frag_failed:
continue
shutil.copyfile(f_img, out_img(img_id))
def main_site_vasegen():
for f_img in tqdm(glob.glob(dir_in + '/*')):
try:
img = dip.imread(f_img)
except:
continue
img_id = int(os.path.split(f_img)[-1].split('.')[0])
# img_out = dip.resize(img, _pix2pix_outsize, interpolation=INTER_AREA)
print(f_img, img.shape)
if len(img.shape) == 3:
gray = np.mean(img, axis=-1)
else:
# gray = img
continue
grad = dip.transforms.edge_detect(gray)
mask, m_min, m_max, n_min, n_max = space_fill(grad)
trimx = grad.shape[0] // 20
trimy = grad.shape[1] // 20
grad[:trimx, :] = 0
grad[-trimx:, :] = 0
grad[:, :trimy] = 0
grad[:, -trimy:] = 0
markerx = _pix2pix_marker_size * (
grad.shape[0]) // _pix2pix_outsize[0] // 2
markery = _pix2pix_marker_size * (
grad.shape[1]) // _pix2pix_outsize[1] // 2
grad_top = grad > np.percentile(grad, 99)
border_inds = np.argwhere(grad_top)
hull = ConvexHull(border_inds)
_frag_context = np.zeros_like(grad_top)
mm = list(np.ndindex(_frag_context.shape[:2]))
out_hull = isInHull(mm, hull)
mm = np.array(mm)[out_hull]
# print(mm)
# input(out_hull)
_frag_context = np.stack([255 * ~_frag_context] * 3 + [_frag_context],
axis=-1).astype(np.uint8)
_frag_context[mm[:, 0], mm[:, 1], :3] = img[mm[:, 0], mm[:, 1]]
_frag_context[mm[:, 0], mm[:, 1], 3] = 255
dip.im_write(_frag_context, out_img(img_id))
def _test_restore_mode(self, file, deg_type, save_images=False, name=None):
fh = ImageFileHandler()
im_degrader = ImageDegrader()
im = fh.open_image_file_as_matrix(file)
degraded_im = im_degrader.degrade(im,
degradation_type=deg_type,
severity_value=.5)
restored_im, clustered_im, h_params = self.multiplicative_clustering_restore(
degraded_im)
restored_im_2 = self.fast_multiplicative_restore(
degraded_im, h_param=np.mean(h_params), search_window_size=21)
if save_images:
dip.im_write(dip.float_to_im(degraded_im),
"./" + name + "_degraded_image.jpg",
quality=95)
dip.im_write(dip.float_to_im(restored_im),
"./" + name + "_restored_image.jpg",
quality=95)
dip.im_write(dip.float_to_im(clustered_im),
"./" + name + "_clustered_image.jpg",
quality=95)
dip.figure()
dip.subplot(141)
dip.imshow(im, cmap="gray")
dip.subplot(142)
dip.imshow(degraded_im, cmap="gray")
dip.xlabel("PSNR: {0:.2f}, SSIM: {1:.2f}".format(
dip.PSNR(im, degraded_im),
dip.SSIM(im, degraded_im)[0]))
dip.subplot(143)
dip.imshow(restored_im, cmap="gray")
dip.xlabel("PSNR: {0:.2f}, SSIM: {1:.2f}".format(
dip.PSNR(im, restored_im),
dip.SSIM(im, restored_im)[0]))
dip.subplot(144)
dip.imshow(restored_im_2, cmap="gray")
dip.xlabel("PSNR: {0:.2f}, SSIM: {1:.2f}".format(
dip.PSNR(im, restored_im_2),
dip.SSIM(im, restored_im_2)[0]))
dip.show()
#(c) Reading an image X = dip.im_read(picture_link) #(d) Converting the image to normalized floating point space X = dip.im_to_float(X) X *= 255 #(e) Adding Constant to Image Y = X + 75 #(f) Renormalize the image and covert to integer Y = dip.float_to_im(Y/255) #(g) Writing an image to disk dip.im_write(Y, save_link_1) #(h) Square Intenstiy and write image to disk Z = X**2 Z = dip.float_to_im(Z/255) Z = dip.im_write(Z, save_link_2) #(i) Compute FFT of X fX = dip.fft2(X) fX = dip.fftshift(fX) fX = np.log(np.abs(fX)) #(j) Save and show the resulting spectrum dip.imshow(fX) dip.show()
# (3) Find the min-area contour
_, cnts, _ = cv2.findContours(threshed, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
cntsSorted = sorted(cnts, key=cv2.contourArea)
cnt = cntsSorted[-1]
# (4) Create mask and do bitwise-op
mask = np.zeros(img.shape[:2], np.uint8)
cv2.drawContours(mask, [cnt], -1, 255, -1)
dst = cv2.bitwise_and(img, img, mask=mask)
# (5) Crop image to JUST around the object, resize so object is in center
x, y, w, h = util.find_bounding_rect(dst)
dstcrop = dst[y:y + h, x:x + w]
dstcrop = util.resize_image(dstcrop, M_max, N_max, 'constant')
# (6) Logical OR - for accumulation of images
accum_object = cv2.bitwise_or(dstcrop, accum_object)
accum_total = cv2.bitwise_or(dstcrop, accum_total)
x, y, w, h = util.find_bounding_rect(accum_object)
accum_object = accum_object[y:y + h, x:x + w]
dip.im_write(accum_object, PATH_OUT_TEMPLATE_IMAGES + 'template_' + str(template_id).zfill(2) + ".jpg")
template_id = template_id + 1
x, y, w, h = util.find_bounding_rect(accum_total)
accum_total = accum_total[y:y + h, x:x + w]
dip.im_write(accum_total, PATH_OUT_TEMPLATE_IMAGES + 'template_master.jpg')
###########################
# RESIZING IMAGES #
###########################
dir_no = "../grayscale_data/no/"
dir_yes = "../grayscale_data/yes/"
dirs = [dir_no, dir_yes]
for directory in dirs:
for filename in os.listdir(directory):
im = dip.im_read(directory + filename)
out = dip.resize(im, (350, 300), interpolation=cv2.INTER_CUBIC)
filename_without_extension = os.path.splitext(filename)[0]
if directory == dir_no:
dip.im_write(out,
"../resized_data/no/" + filename_without_extension +
".png",
quality=95) # 95 is the best possible image quality
elif directory == dir_yes:
dip.im_write(out,
"../resized_data/yes/" + filename_without_extension +
".png",
quality=95) # 95 is the best possible image quality
###########################
# VARYING CONTRAST #
###########################
dir_no = "../resized_data/no/"
dir_yes = "../resized_data/yes/"
dirs = [dir_no, dir_yes]
for directory in dirs:
def save_matrix_as_image_file(self, matrix, image_path):
dip.im_write(dip.float_to_im(matrix), image_path)
## Change the path to where you want to save the resulting images
path_mix = "/Users/chuchu/Dropbox/gt_exp/set2/image_mix/"
##
o_c_e = [(1, 0, 0), (0, 1, 0), (0, 0, 1), (0.5, 0.25, 0.25), (0.25, 0.5, 0.25),
(0.25, 0.25, 0.5), (0.33, 0.33, 0.33)]
dir = path_mix
if not os.path.exists(dir):
os.makedirs(dir)
for event in range(3, len(o_c_e)):
dir = path_mix + 'event_' + str(o_c_e[event][0]) + '_' + str(
o_c_e[event][1]) + '_' + str(o_c_e[event][2])
if not os.path.exists(dir):
os.makedirs(dir)
for filename in os.listdir(path_orgin):
path_image_origin = path_orgin + filename
path_image_contrast = path_contrast + filename
path_image_edge = path_edge + filename
im_origin = dip.im_read(path_image_origin) # read icmage
im_contrast = dip.im_read(path_image_contrast) # read icmage
im_edge = dip.im_read(path_image_edge) # read icmage
im_mix = (im_origin * o_c_e[event][0] + im_contrast * o_c_e[event][1] +
im_edge * o_c_e[event][2]).astype(np.uint8)
save_name = dir + '/' + filename[:-4] + '.png'
dip.im_write(im_mix, save_name)
import numpy as np
import matplotlib.pyplot as plt
import os
path = "/Users/chuchu/Dropbox/gt_exp/ddb1_fundusimages/" # Change the path to where you store the images
modes = [2, 3, 4]
for m in range(len(modes)):
dir = 'upsample_img/'+str(modes[m]);
if not os.path.exists(dir):
os.makedirs(dir)
for m in range(len(modes)):
for filename in os.listdir(path):
mode = modes[m]
file_path = path + filename
f = dip.im_read(file_path) # read icmage
f = dip.im_to_float(f)
M = np.array([[1/mode, 0],
[0, 1/mode]])
f_up_0 = dip.sampling.resample(f[:,:,0], M, interp= 'bilinear')
f_up_1 = dip.sampling.resample(f[:, :, 1], M, interp='bilinear')
f_up_2 = dip.sampling.resample(f[:, :, 2], M, interp='bilinear')
h, l = f_up_0.shape
f_up = np.zeros((h, l, 3))
f_up[:,:,0] = f_up_0
f_up[:, :, 1] = f_up_1
f_up[:, :, 2] = f_up_2
upsample_img = 'upsample_img/' + str(mode)+ '/' +filename[:-4] + '_up' + str(mode) + '.jpg'
save_im = dip.float_to_im(f_up)
dip.im_write(save_im, upsample_img)
import dippykit as dip
from cv2 import INTER_AREA, INTER_CUBIC
try:
outsize = (int(sys.argv[1]) * 2, int(sys.argv[1]))
except:
print('provide a number for the side of the square image to scale to')
exit()
_pix2pix_indir = 'data/processed/pix2pix_vase_fragments/train/'
_pix2pix_outdir = f'data/processed/pix2pix_vase_fragments_{sys.argv[1]}/train/'
out_pix2pix = lambda fname: f'{_pix2pix_outdir}/{fname}.jpg'
if os.path.exists(_pix2pix_outdir):
y_n = input(f'Folder {_pix2pix_outdir} exists, overwrite?')
if 'y' not in y_n:
exit()
os.makedirs(_pix2pix_outdir, exist_ok=True)
for f_img in tqdm(glob.glob(_pix2pix_indir + '/*.jpg')):
try:
img = dip.imread(f_img)
except:
continue
img_name = os.path.split(f_img)[-1]
img = dip.resize(img, outsize, interpolation=INTER_CUBIC)
dip.im_write(img, out_pix2pix(img_name))
dir = 'enhance/';
if not os.path.exists(dir):
os.makedirs(dir)
for mode in modes:
for filename in os.listdir(path):
file_path = path + filename
f = dip.im_read(file_path) #read icmage
f = dip.im_to_float(f)
f_g = rgb2gray(f)
f_edge = dip.transforms.edge_detect(f_g, mode = mode, as_bool=False);
edgename = 'edge/' + filename[:-4] + '.png';
if mode is not 'canny':
f_edge *= 1/f_edge.max()
save_im = dip.float_to_im(f_edge)
dip.im_write(save_im, edgename)
f[:,:,0] += f_edge
f[:, :, 1] += f_edge
f[:, :, 2] += f_edge
enhance_img = 'enhance/' + filename[:-4] + '.png';
save_im = dip.float_to_im(f)
dip.im_write(save_im, enhance_img)
####combine all
#
# for filename in os.listdir(path):
# file_path = path + filename
# f = dip.im_read(file_path) # read icmage
# f = dip.im_to_float(f)
# f_g = rgb2gray(f)
def main_pix2pix_context():
def out_pix2pix_name():
global _pix2pix_counter
outname = f'{_pix2pix_dir}/{_pix2pix_counter}.jpg'
_pix2pix_counter += 1
return outname
for f_img in tqdm(glob.glob(dir_in + '/*')):
try:
img = dip.imread(f_img)
except:
continue
img_id = int(os.path.split(f_img)[-1].split('.')[0])
# img_out = dip.resize(img, _pix2pix_outsize, interpolation=INTER_AREA)
print(f_img, img.shape)
if len(img.shape) == 3:
gray = np.mean(img, axis=-1)
else:
# gray = img
continue
grad = dip.transforms.edge_detect(gray)
mask, m_min, m_max, n_min, n_max = space_fill(grad)
trimx = grad.shape[0] // 20
trimy = grad.shape[1] // 20
grad[:trimx, :] = 0
grad[-trimx:, :] = 0
grad[:, :trimy] = 0
grad[:, -trimy:] = 0
markerx = _pix2pix_marker_size * (grad.shape[0]) // _pix2pix_outsize[0] // 2
markery = _pix2pix_marker_size * (grad.shape[1]) // _pix2pix_outsize[1] // 2
grad_top = grad > np.percentile(grad, 99)
border_inds = np.argwhere(grad_top)
hull = ConvexHull(border_inds)
perim = border_inds[hull.vertices]
perim = np.concatenate((perim, border_inds[hull.vertices[:1]]), axis=0)
_frag_context = np.zeros_like(grad_top)
for a, b in zip(perim[:-1], perim[1:]):
a, b = np.array(a), np.array(b)
for t in np.linspace(0, 1, max(grad.shape)):
p = t * a + (1 - t) * b
ind = np.round(p).astype(np.int)
mstart = max(ind[0] - markerx, 0)
mstop = min(ind[0] + markerx, grad.shape[0] - 1)
nstart = max(ind[1] - markery, 0)
nstop = min(ind[1] + markery, grad.shape[1] - 1)
_frag_context[mstart:mstop, nstart:nstop] = 1
mm = list(np.ndindex(_frag_context.shape[:2]))
out_hull = isInHull(mm, hull)
mm = np.array(mm)[~out_hull]
# print(mm)
# input(out_hull)
_frag_context = np.stack([255*~_frag_context]*3, axis=-1).astype(np.uint8)
_frag_context[mm[:, 0], mm[:, 1]] = img[mm[:, 0], mm[:, 1]]
# binary erosion - erodes binary blocks like sand bars
# frag_context2 = frag_context & binary_erosion(frag_context)
# threshold/contour method
# ret, thresh = cv2.threshold(grad, np.percentile(grad, 95), 255, 0)
# contours, hier = cv2.findContours(thresh.astype(np.uint8), cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
# print(contours, hier.shape)
# frag_context = np.zeros_like(grad)
# cv2.drawContours(frag_context, contours, -1, 255, 3)
# draw contours according to hierarchy
# frag_context2 = np.zeros_like(grad)
# hier = (hier - np.min(hier)) / (np.max(hier) - np.min(hier))
# for h, contlist in zip(hier[0], contours):
# for cont in contlist:
# frag_context2[cont[0, 1], cont[0, 0]] = max(h)
# otsu thresholding
# val = filters.threshold_otsu(gray); frag_context = gray < val
for n in range(n_fragments):
# frag_size = max(img.shape[0], img.shape[1]) // 4
frag_size = img.shape[0]//4, img.shape[1]//4
frag, frag_pos = fragment(img, m_min, m_max, n_min, n_max, frag_size=frag_size)
fragx = slice(frag_pos[0], frag_pos[0]+frag_size[0])
fragy = slice(frag_pos[1], frag_pos[1]+frag_size[1])
if frag is None:
break
else:
frag_context = np.copy(_frag_context)
white = (255, 255, 255)
frag_context[fragx, fragy][frag != white] = frag[frag != white]
# img = np.copy(img)
# img[fragx, fragy][frag != white] = frag[frag != white]
# frag_context_out = dip.resize(frag_context, _pix2pix_outsize, interpolation=INTER_AREA)
# frag_context_out = frag_context_out[:, :, 0]
# print(img_out.shape, frag_context_out.shape)
# print(img_out.dtype, frag_context_out.dtype)
# both = np.concatenate([img_out, frag_context_out], axis=1)
both = np.concatenate([img, frag_context], axis=1)
# plt.subplot(221)
# plt.imshow(img)
# plt.subplot(222)
# plt.imshow(grad)
# plt.subplot(223)
# plt.imshow(frag_context)
# plt.subplot(224)
# plt.imshow(both)
# plt.show()
dip.im_write(both, out_pix2pix(img_id, n))
