def labnames2imgs_cens(labnames, n):
data = []
for ln in labnames:
t, z = (int(ln[-11:-8]), int(ln[-7:-4]))
z2 = floor((z + 0.5) * 5.5)
img = imread(imgnames(n)[t])
cen = imread(cennames(n)[t])
lab = imread(ln)
lab = zoom(lab, (0.5, 0.5), order=0)
lab = label(lab)[0]
img = gputools.scale(img, (5.5, 0.5, 0.5))
img = img[..., np.newaxis]
## upscale centerpoints. make them just a single pixel.
pts = np.array([n['centroid']
for n in nhl_tools.hyp2nhl(cen)]).astype(np.int)
pts = np.floor((pts + 0.5) * [5.5, 0.5, 0.5]).astype(np.int)
cen = np.zeros(img.shape[:-1])
cen[tuple(pts.T)] = 1
data.append({
'img': img,
'cen': cen,
'lab': lab,
't': t,
'z': z,
'z2': z2,
'labname': ln
})
return data
def build_rawdata(homedir, savedir):
homedir = Path(homedir)
savedir = Path(savedir)
labnames = glob(str(homedir / 'Fluo-N3DH-CE/01_GT/SEG/man_seg_*'))
print(labnames)
cat = np.concatenate
def imread2(x): return imread(str(x))
imgs = []
cens = []
labs = []
times_zs = []
hyps = []
for i in range(len(labnames)):
i = 2
ln = labnames[i]
tz = (int(ln[-11:-8]), int(ln[-7:-4]))
t,z = tz
lab = imread2(ln)
img = imread2(homedir / 'Fluo-N3DH-CE/01/t{:03d}.tif'.format(t))
cen = imread2(homedir / 'Fluo-N3DH-CE/01_GT/TRA/man_track{:03d}.tif'.format(t))
img = gputools.scale(img, (5.5,0.5,0.5))
lab = zoom(lab, (0.5,0.5), order=0)
lab = label(lab)[0]
# cen = label(cen)[0].astype(np.uint16)
cen = cen>0
cen = gputools.scale(cen, (5.5,0.5,0.5), interpolation='linear')
cen = label(cen>0.5)[0]
img2 = gputools.denoise.nlm3(img, sigma=20)
hyp = watershed(img2, cen, mask=img2>img2.mean()*1.0, compactness=1000)
z2 = int(z*5.5)
print(ln)
print(scores_dense.seg(lab, hyp[z2]))
# ipdb.set_trace()
return locals()
def regionotsu(img3):
# threshimg = np.ones(img3.shape)
sh = (50,1,2)
threshimg = np.zeros(sh)
patchshape = [ceil(1*img3.shape[i]/sh[i]) for i in range(img3.ndim)]
# res = patchmaker.patchtool({'img':img3.shape,'patch':(170//2,256//4,356//4),'overlap_factor':(2,2,2)})
res = patchmaker.patchtool({'grid':sh,'img':img3.shape,'patch':patchshape})
for i in range(len(res['starts'])):
ss = res['slices'][i]
ind = res['inds'][i]
th = otsu3d(img3[ss])
# th = np.percentile(img3[ss],90) #*0.75 #.mean()*2.5
# th = img3[ss].max
threshimg[tuple(ind)] = th
scale = [img3.shape[i]/threshimg.shape[i] for i in range(img3.ndim)]
threshimg = gputools.scale(threshimg.astype(np.float32),scale)
return threshimg
def test_scale():
d = np.zeros((100,100,100),np.float32)
scale(d,1.7, interp = "linear")
scale(d,.3, interp = "linear")
def test_scale():
d = np.zeros((100, 100, 100), np.float32)
scale(d, 1.7, interp="linear")
scale(d, .3, interp="linear")
def test_scale(x):
s = np.random.uniform(.5,1.5,3)
out1 = scale(x, s, interpolation="nearest")
out2 = ndimage.zoom(x, s, order=0, prefilter=False)
return out1, out2
def crop_stack(data, sc=(5.5, .5, .5)):
newshape = _scale_shape(_scale_shape(data.shape, sc),
tuple(1. / s for s in sc))
return pad_to_shape(data, newshape)
def pad_stack(data, target_shape):
return pad_to_shape(data, target_shape)
data_target = np.zeros((31, 512, 712), np.uint16)
sc = (5.5, .5, .5)
sc_inv = tuple(1. / s for s in sc)
data_crop = crop_stack(data_target, sc)
data_crop_scaled = scale(data_crop, sc)
# do segmenting on data_crop_scaled, resulting in seg_crop_scaled
seg_crop_scaled = data_crop_scaled
seg_crop = scale(seg_crop_scaled, sc_inv)
seg_target = pad_stack(seg_crop, data_target.shape)
print(data_target.shape, seg_target.shape)
def test_scale():
d = create_shape((101,102,103))
scale(d,1.7, interp = "linear")
out = scale(d,.3, interp = "nearest")
return out
def f(idx):
return gputools.scale(img[idx[0], ..., idx[1]], scale)
def build_rawdata(homedir):
homedir = Path(homedir)
## load data
# img = imread(str(homedir / 'data/img006.tif'))
img = np.load(str(homedir / 'data/img006_noconv.npy'))
img = img[1]
img = perm(img, "ZCYX", "ZYXC")
img = norm_szyxc_per(img, (0, 1, 2))
img = img.astype(np.float32)
points = lib.mkpoints()
imgsem = {'axes': "ZYXC", 'nuc': 1, 'mem': 0, 'n_channels': 2}
# img = img[:70,::2,::2]
scale = (2.5, .5, .5)
def f(idx):
return gputools.scale(img[..., idx[0]], scale)
img = ts.broadcast_over(f, (2, ))
img = perm(img, "czyx", "zyxc")
points = (points * scale).astype(np.int)
# points[:,0] = points[:,0].clip(0,img.shape[0]-1)
# scale!
# img = zoom(img, (5,1,1,1))
if False:
shape = np.array(img.shape)
res = patchmaker.patchtool({
'img': shape,
'patch': shape // 2,
'grid': (2, 2, 2, 2)
})
scaled = [
gputools.scale((img[ss][..., 0]).astype(np.float32), (5, 1, 1))
for ss in res['slices']
]
ends = res['starts'] + shape // 2 * [5, 1, 1, 1]
container = np.zeros(shape * [5, 1, 1, 1])
big_patches = patchmaker.starts_ends_to_slices(res['starts'], ends)
for i in range(len(big_patches)):
container[big_patches[i]] = scaled[i][..., np.newaxis]
container = img.copy()
cen = np.zeros(container.shape[:-1])
cen[list(points.T)] = 1
x = container[..., 1]
hx = np.array([1, 1, 1]) / 3
x = gputools.convolve_sep3(x, hx, hx, hx, sub_blocks=(1, 1, 1))
lab = watershed(-x, label(cen)[0], mask=x > x.mean())
# bor = np.array([morph.binary_dilation(b) for b in bor]) ## just x,y
# bor = np.array([morph.binary_dilation(b) for b in bor]) ## just x,y
if False:
target = lab[::2, ::2, ::2]
points = points // 2
bg = target == 0
fg = target == 1
target[fg] = 0.0
hx = np.array([1, 1, 1]) / 3
for i in range(200):
target[bg] = 0
target[list(points.T)] = 1
target = gputools.convolve_sep3(target,
hx,
hx,
hx,
sub_blocks=(2, 2, 2))
points = points * 2
target = gputools.scale(target, (2, 2, 2))
if False:
xx = np.linspace(-2, 2, 80)
xx = np.exp(-xx**2 / 2)
xx = xx / xx.sum()
gauss = gputools.convolve_sep3(cen, xx, xx, xx, sub_blocks=(2, 2, 2))
if True:
target = lab2bgdist(lab)
if False:
target = target * gauss
target = np.exp(-target / 10)
res = dict()
res['source'] = container
res['imgsem'] = imgsem
res['target'] = target
res['points'] = points
res['cen'] = cen
return res
def test_scale():
d = create_shape((101, 102, 103))
scale(d, 1.7, interpolation="linear")
out = scale(d, .3, interpolation="nearest")
return out
def f(x):
# t,c = ind
# x = arr_in[t,c]
x = gputools.scale(x, (subsample, 1, 1), interpolation='linear')
return x
