def run_algo(self, stdout=None):
"""
The core algo runner
could also be called by a batch processor
this one needs no parameter
"""
timeout = False
# Get image size
size = image(self.work_dir + 'input_0.png').size
# Run tvrestore
self.wait_proc(
self.run_proc([
'fstv', 'u0.png', 'mask.png', 'inpainted.png',
str(self.cfg['param']['alpha']),
str(self.cfg['param']['beta'])
],
stdout=stdout,
stderr=stdout), timeout)
zoomfactor = int(max(1, floor(550.0 / max(size[0], size[1]))))
size = (zoomfactor * size[0], zoomfactor * size[1])
for filename in ['input_0', 'composite', 'inpainted']:
img = image(self.work_dir + filename + '.png')
img.resize(size, method='nearest')
img.save(self.work_dir + filename + '_zoom.png')
self.cfg['param']['displayheight'] = max(200, size[1])
self.cfg['param']['zoomfactor'] = zoomfactor
self.cfg['param']['zoomwidth'] = size[0]
self.cfg['param']['zoomheight'] = size[1]
def select_subimage(self, x0, y0, x1, y1):
"""
cut subimage from original image
"""
# draw selected rectangle on the image
imgS = image(self.work_dir + 'input_0.png')
imgS.draw_line([(x0, y0), (x1, y0), (x1, y1), (x0, y1), (x0, y0)],
color="red")
imgS.draw_line([(x0+1, y0+1), (x1-1, y0+1), (x1-1, y1-1), (x0+1, y1-1),
(x0+1, y0+1)], color="white")
imgS.save(self.work_dir + 'input_0s.png')
# crop the image
# try cropping from the original input image (if different from input_0)
im0 = image(self.work_dir + 'input_0.orig.png')
dx0 = im0.size[0]
img = image(self.work_dir + 'input_0.png')
dx = img.size[0]
if (dx != dx0) :
z = float(dx0)/float(dx)
im0.crop((int(x0*z), int(y0*z), int(x1*z), int(y1*z)))
# resize if cropped image is too big
if self.input_max_pixels and prod(im0.size) > self.input_max_pixels:
im0.resize(self.input_max_pixels, method="antialias")
img = im0
else :
img.crop((x0, y0, x1, y1))
# save result
img.save(self.work_dir + 'input_0.sel.png')
return
def wait(self, **kwargs):
"""
params handling and run redirection
"""
if 'precision' in kwargs:
self.cfg['param']['precision'] = kwargs['precision']
if 'SiftRatio' in kwargs:
self.cfg['param']['siftratio'] = kwargs['SiftRatio']
im0 = 'input_0.png'
if 'rectangle' in kwargs:
try:
self.cfg['param']['x0'] = int(kwargs['x0'])
self.cfg['param']['y0'] = int(kwargs['y0'])
self.cfg['param']['x1'] = int(kwargs['x1'])
self.cfg['param']['y1'] = int(kwargs['y1'])
except ValueError:
return self.error(errcode='badparams',
errmsg="Incorrect parameters.")
im0 = 'input_crop.png'
# no parameter
http.refresh(self.base_url + 'run?key=%s' % self.key)
width = max(image(self.work_dir + im0).size[0],
image(self.work_dir + 'input_1.png').size[0])
height = max(image(self.work_dir + im0).size[1],
image(self.work_dir + 'input_1.png').size[1])
return self.tmpl_out("wait.html", width=width, height=height, im0=im0)
def select_subimage(self, x0, y0, x1, y1):
"""
cut subimage from original image
"""
# draw selected rectangle on the image
imgS = image(self.work_dir + 'input_0.png')
imgS.draw_line([(x0, y0), (x1, y0), (x1, y1), (x0, y1), (x0, y0)],
color="red")
imgS.draw_line([(x0 + 1, y0 + 1), (x1 - 1, y0 + 1), (x1 - 1, y1 - 1),
(x0 + 1, y1 - 1), (x0 + 1, y0 + 1)],
color="white")
imgS.save(self.work_dir + 'input_0s.png')
# crop the image
# try cropping from the original input image
# (if different from input_1)
im0 = image(self.work_dir + 'input_0.orig.png')
dx0 = im0.size[0]
img = image(self.work_dir + 'input_0.png')
dx = img.size[0]
if (dx != dx0):
z = float(dx0) / float(dx)
im0.crop((int(x0 * z), int(y0 * z), int(x1 * z), int(y1 * z)))
# resize if cropped image is too big
if (self.input_max_pixels
and prod(im0.size) > self.input_max_pixels):
im0.resize(self.input_max_pixels, method="antialias")
img = im0
else:
img.crop((x0, y0, x1, y1))
# save result
img.save(self.work_dir + 'input_0.sel.png')
return
def run_algo(self, s, k, t, stdout=None):
"""
the core algo runner
could also be called by a batch processor
this one needs no parameter
"""
p = self.run_proc(['psf_estim',
'-p', self.bin_dir + 'pattern_noise.pgm',
'-s', str(s),
'-k', str(k),
'-d', 'det_out.ppm',
'-t', str(t),
'-o', 'psf.pgm',
'input_0.pgm', 'psf.txt'],
stdout=stdout, stderr=stdout)
self.wait_proc(p)
im = image(self.work_dir + "psf.pgm")
# re adjust width and height to avoid visualization interpolation
width = 600
height = 600
# interpolate it by neareset neighbor
im = im.resize((width, height), "nearest")
im.save(self.work_dir + "psf.png")
im = image(self.work_dir + "det_out.ppm")
im.save(self.work_dir + "det_out.png")
return
def run_algo(self, stdout=None):
"""
The core algo runner
could also be called by a batch processor
this one needs no parameter
"""
timeout = False
files = ['input_0_sel', 'detectedraw', 'ppSkin', 'ppRobust', 'ppBoth']
self.wait_proc(self.run_proc(['detect', 'input_0_sel.png', str(self.cfg['param']['layercount']), str(self.cfg['param']['threshold'])],
stdout=stdout, stderr=stdout), timeout*0.90)
# Resize for visualization (always zoom by at least 2x)
(sizeX, sizeY) = image(self.work_dir + 'input_0_sel.png').size
zoomfactor = max(1, int(math.ceil(480.0/max(sizeX, sizeY))))
if zoomfactor > 1:
(sizeX, sizeY) = (zoomfactor*sizeX, zoomfactor*sizeY)
for filename in files:
im = image(self.work_dir + filename + '.png')
im.resize((sizeX, sizeY), method='nearest')
im.save(self.work_dir + filename + '_zoom.png')
self.cfg['param']['disp_suffix'] = '_zoom.png'
else:
self.cfg['param']['disp_suffix'] = '.png'
self.cfg['param']['zoomfactor'] = zoomfactor
self.cfg['param']['displayheight'] = max(200, sizeY)
self.cfg['param']['stdout'] = \
open(self.work_dir + 'stdout.txt', 'r').read()
def params(self, newrun=False, msg=None):
"""
configure the algo execution
"""
if newrun:
old_work_dir = self.work_dir
old_cfg_param = self.cfg['param']
self.clone_input()
self.cfg['param'].update(old_cfg_param)
self.cfg['param']['norectif'] = True # No rectif for new run
self.cfg.save()
cp_list = ['H_input_0.png', 'H_input_1.png']
for fname in cp_list:
shutil.copy(old_work_dir + fname, self.work_dir + fname)
if 'radius' not in self.cfg['param']:
self.cfg['param']['radius'] = 9
self.cfg['param']['alpha'] = 0.9
self.cfg.save()
if 'sense' not in self.cfg['param']:
self.cfg['param']['sense'] = 'r'
if (image(self.work_dir + 'input_0.png').size !=
image(self.work_dir + 'input_1.png').size):
return self.error('badparams',
"The images must have the same size")
return self.tmpl_out("params.html")
def run_algo(self, stdout=None):
"""
The core algo runner
could also be called by a batch processor
this one needs no parameter
"""
timeout = False
# Get image size
size = image(self.work_dir + 'input_0.png').size
# Run tvrestore
self.wait_proc(self.run_proc(['tvinpaint', 'mask.png',
str(self.cfg['param']['lambda']), 'u0.png', 'inpainted.png'],
stdout=stdout, stderr=stdout), timeout)
zoomfactor = int(max(1, floor(550.0/max(size[0], size[1]))))
size = (zoomfactor*size[0], zoomfactor*size[1])
for filename in ['input_0', 'composite', 'inpainted']:
img = image(self.work_dir + filename + '.png')
img.resize(size, method='nearest')
img.save(self.work_dir + filename + '_zoom.png')
self.cfg['param']['displayheight'] = max(200, size[1])
self.cfg['param']['zoomfactor'] = zoomfactor
self.cfg['param']['zoomwidth'] = size[0]
self.cfg['param']['zoomheight'] = size[1]
def run_algo(self, stdout=None, timeout=False):
"""
The core algo runner
could also be called by a batch processor
this one needs no parameter
"""
# Get image size
size = image(self.work_dir + 'input_0.png').size
# Run eple: EPLE is allowed to iterate just once because of the its speed
# PLE, even slower, isn't fit for the demo.
self.wait_proc(self.run_proc(['inpaintEPLE', 'u0.png', 'mask.png',
'inpainted.png', str(self.cfg['param']['iteration']) ], stdout=stdout, stderr=stdout), timeout)
zoomfactor = int(max(1, floor(550.0/max(size[0], size[1]))))
size = (zoomfactor*size[0], zoomfactor*size[1])
for filename in ['input_0', 'composite', 'inpainted']:
img = image(self.work_dir + filename + '.png')
img.resize(size, method='nearest')
img.save(self.work_dir + filename + '_zoom.png')
self.cfg['param']['displayheight'] = max(200, size[1])
self.cfg['param']['zoomfactor'] = zoomfactor
self.cfg['param']['zoomwidth'] = size[0]
self.cfg['param']['zoomheight'] = size[1]
self.cfg['param']['stdout'] = \
open(self.work_dir + 'stdout.txt', 'r').read()
def run_algo(self, s, k, t, stdout=None):
"""
the core algo runner
could also be called by a batch processor
this one needs no parameter
"""
p = self.run_proc([
'psf_estim', '-p', self.bin_dir + 'pattern_noise.pgm', '-s',
str(s), '-k',
str(k), '-d', 'det_out.ppm', '-t',
str(t), '-o', 'psf.pgm', 'input_0.pgm', 'psf.txt'
],
stdout=stdout,
stderr=stdout)
self.wait_proc(p)
im = image(self.work_dir + "psf.pgm")
# re adjust width and height to avoid visualization interpolation
width = 600
height = 600
# interpolate it by neareset neighbor
im = im.resize((width, height), "nearest")
im.save(self.work_dir + "psf.png")
im = image(self.work_dir + "det_out.ppm")
im.save(self.work_dir + "det_out.png")
return
def run_algo(self, stdout=None):
"""
The core algo runner
could also be called by a batch processor
this one needs no parameter
"""
timeout = False
pattern = self.cfg['param']['pattern']
# Mosaic image
self.wait_proc(
self.run_proc(
['mosaic', '-p', pattern, 'input_0_sel.png', 'mosaiced.png'],
stdout=stdout,
stderr=stdout), timeout * 0.2)
# Demosaic image
self.wait_proc([
self.run_proc(
['dmzhangwu', '-p', pattern, 'mosaiced.png', 'dmzhangwu.png'],
stdout=stdout,
stderr=stdout),
self.run_proc(
['dmbilinear', '-p', pattern, 'mosaiced.png', 'bilinear.png'],
stdout=stdout,
stderr=stdout)
], timeout * 0.6)
# Compute image differences
self.wait_proc([
self.run_proc([
'imdiff', 'input_0_sel.png', 'dmzhangwu.png',
'diffdmzhangwu.png'
],
stdout=stdout,
stderr=stdout),
self.run_proc([
'imdiff', 'input_0_sel.png', 'bilinear.png', 'diffbilinear.png'
],
stdout=stdout,
stderr=stdout)
], timeout * 0.2)
# Resize for visualization (new size of the smallest dimension = 200)
(sizeX, sizeY) = image(self.work_dir + 'input_0_sel.png').size
zoomfactor = max(1, int(ceil(200.0 / min(sizeX, sizeY))))
(sizeX, sizeY) = (zoomfactor * sizeX, zoomfactor * sizeY)
for filename in [
'input_0_sel', 'mosaiced', 'dmzhangwu', 'bilinear',
'diffdmzhangwu', 'diffbilinear'
]:
im = image(self.work_dir + filename + '.png')
im.resize((sizeX, sizeY), method='nearest')
im.save(self.work_dir + filename + '_zoom.png')
self.cfg['param']['displayheight'] = sizeY
self.cfg['param']['zoomfactor'] = zoomfactor
def result(self):
"""
display the algo results
"""
print "Display results"
# read the parameters
sigma = self.cfg['param']['sigma']
try:
x0 = self.cfg['param']['x0']
except KeyError:
x0 = None
try:
y0 = self.cfg['param']['y0']
except KeyError:
y0 = None
try:
x1 = self.cfg['param']['x1']
except KeyError:
x1 = None
try:
y1 = self.cfg['param']['y1']
except KeyError:
y1 = None
(sizeX, sizeY) = image(self.work_dir + 'input_0.png').size
# Resize for visualization (new size of the smallest dimension = 200)
zoom_factor = None
if (sizeX < 200) or (sizeY < 200):
if sizeX > sizeY:
zoom_factor = int(ceil(200.0/sizeY))
else:
zoom_factor = int(ceil(200.0/sizeX))
sizeX = sizeX*zoom_factor
sizeY = sizeY*zoom_factor
im = image(self.work_dir + 'input_0.png')
im.resize((sizeX, sizeY), method="pixeldup")
im.save(self.work_dir + 'input_0_zoom.png')
im = image(self.work_dir + 'input_1.png')
im.resize((sizeX, sizeY), method="pixeldup")
im.save(self.work_dir + 'input_1_zoom.png')
im = image(self.work_dir + 'output_1.png')
im.resize((sizeX, sizeY), method="pixeldup")
im.save(self.work_dir + 'output_1_zoom.png')
im = image(self.work_dir + 'output_2.png')
im.resize((sizeX, sizeY), method="pixeldup")
im.save(self.work_dir + 'output_2_zoom.png')
return self.tmpl_out("result.html", sigma=sigma,
x0=x0, y0=y0, x1=x1, y1=y1,
sizeY=sizeY, zoom_factor=zoom_factor)
def result(self):
"""
display the algo results
"""
# read the parameters
scale = self.cfg['param']['scale']
try:
x0 = self.cfg['param']['x0']
except KeyError:
x0 = None
try:
y0 = self.cfg['param']['y0']
except KeyError:
y0 = None
try:
x1 = self.cfg['param']['x1']
except KeyError:
x1 = None
try:
y1 = self.cfg['param']['y1']
except KeyError:
y1 = None
(sizeX, sizeY) = image(self.work_dir + 'input_0.sel.png').size
# Resize for visualization (new size of the smallest dimension = 200)
zoom_factor = None
if (sizeX < 200) or (sizeY < 200):
if sizeX > sizeY:
zoom_factor = int(ceil(200.0 / sizeY))
else:
zoom_factor = int(ceil(200.0 / sizeX))
sizeX = sizeX * zoom_factor
sizeY = sizeY * zoom_factor
im = image(self.work_dir + 'input_0.sel.png')
im.resize((sizeX, sizeY), method="pixeldup")
im.save(self.work_dir + 'input_0_zoom.sel.png')
im = image(self.work_dir + 'cartoon.png')
im.resize((sizeX, sizeY), method="pixeldup")
im.save(self.work_dir + 'cartoon_zoom.png')
im = image(self.work_dir + 'texture.png')
im.resize((sizeX, sizeY), method="pixeldup")
im.save(self.work_dir + 'texture_zoom.png')
return self.tmpl_out("result.html",
scale=scale,
x0=x0,
y0=y0,
x1=x1,
y1=y1,
sizeY=sizeY,
zoom_factor=zoom_factor)
def result(self, error_nomatch=None):
"""
display the algo results
"""
height = max(image(self.work_dir + 'input_0.png').size[1],
image(self.work_dir + 'input_1.png').size[1])
if error_nomatch:
return self.tmpl_out("result_nomatch.html", height=height)
else:
return self.tmpl_out("result.html", height=height)
def params(self, newrun=False, msg=None):
"""
configure the algo execution
"""
if newrun:
self.clone_input()
if (image(self.work_dir + 'input_0.png').size
!= image(self.work_dir + 'input_1.png').size):
return self.error('badparams',
"The images must have the same size")
return self.tmpl_out("params.html")
def result(self, public=None):
"""
display the algo results
"""
resultHeight = image(self.work_dir + 'input_0_selection.png').size[1]
imageHeightResized = min (600, resultHeight)
resultHeight = max(200, resultHeight)
return self.tmpl_out("result.html", height=resultHeight, \
heightImageDisplay=imageHeightResized, \
width=image(self.work_dir\
+'input_0_selection.png').size[0])
def params(self, newrun=False, msg=None):
"""
configure the algo execution
"""
if newrun:
self.clone_input()
width = max(image(self.work_dir + 'input_0.png').size[0],
image(self.work_dir + 'input_1.png').size[0])
height = max(image(self.work_dir + 'input_0.png').size[1],
image(self.work_dir + 'input_1.png').size[1])
return self.tmpl_out("params.html", width=width, height=height)
def run_algo(self, stdout=None):
"""
The core algo runner
could also be called by a batch processor
this one needs no parameter
"""
timeout = False
pattern = self.cfg["param"]["pattern"]
# Mosaic image
self.wait_proc(
self.run_proc(["mosaic", "-p", pattern, "input_0_sel.png", "mosaiced.png"], stdout=stdout, stderr=stdout),
timeout * 0.2,
)
# Demosaic image
self.wait_proc(
[
self.run_proc(
["dmzhangwu", "-p", pattern, "mosaiced.png", "dmzhangwu.png"], stdout=stdout, stderr=stdout
),
self.run_proc(
["dmbilinear", "-p", pattern, "mosaiced.png", "bilinear.png"], stdout=stdout, stderr=stdout
),
],
timeout * 0.6,
)
# Compute image differences
self.wait_proc(
[
self.run_proc(
["imdiff", "input_0_sel.png", "dmzhangwu.png", "diffdmzhangwu.png"], stdout=stdout, stderr=stdout
),
self.run_proc(
["imdiff", "input_0_sel.png", "bilinear.png", "diffbilinear.png"], stdout=stdout, stderr=stdout
),
],
timeout * 0.2,
)
# Resize for visualization (new size of the smallest dimension = 200)
(sizeX, sizeY) = image(self.work_dir + "input_0_sel.png").size
zoomfactor = max(1, int(ceil(200.0 / min(sizeX, sizeY))))
(sizeX, sizeY) = (zoomfactor * sizeX, zoomfactor * sizeY)
for filename in ["input_0_sel", "mosaiced", "dmzhangwu", "bilinear", "diffdmzhangwu", "diffbilinear"]:
im = image(self.work_dir + filename + ".png")
im.resize((sizeX, sizeY), method="nearest")
im.save(self.work_dir + filename + "_zoom.png")
self.cfg["param"]["displayheight"] = sizeY
self.cfg["param"]["zoomfactor"] = zoomfactor
def result(self, public=None):
"""
display the algo results
"""
resultHeight = image(self.work_dir + 'input_0.png').size[1]
imageHeightResized = min(600, resultHeight)
resultHeight = max(200, resultHeight)
return self.tmpl_out("result.html", height=resultHeight, \
heightImageDisplay=imageHeightResized, \
width=image(self.work_dir\
+'input_0.png').size[0])
def result(self):
"""
display the algo results
"""
# read the parameters
L = self.cfg['param']['l']
s = self.cfg['param']['s']
sizeX = image(self.work_dir + 'input_0.png').size[0]
sizeY = max(
image(self.work_dir + 'input_0.png').size[1], 3 * 128 + 4 * 20)
return self.tmpl_out("result.html", L=L, s=s, sizeX=sizeX, sizeY=sizeY)
def run_algo(self, r, stdout=None, timeout=False):
"""
the core algo runner
could also be called by a batch processor
this one needs no parameter
"""
#color correction of R, G and B channels
option = 1
p1 = self.run_proc(['localcolorcorrection', 'input_0.sel.png',
'output_1.png', str(r), str(option)],
stdout=None, stderr=None)
#color correction of intensity channel (keep R/G/B ratios)
option = 2
p2 = self.run_proc(['localcolorcorrection', 'input_0.sel.png',
'output_2.png', str(r), str(option)],
stdout=None, stderr=None)
#color correction of Y channel (use YPbPr color space)
option = 3
p3 = self.run_proc(['localcolorcorrection', 'input_0.sel.png',
'output_3.png', str(r), str(option)],
stdout=None, stderr=None)
#color correction of L channel (use HSL color space)
option = 4
p4 = self.run_proc(['localcolorcorrection', 'input_0.sel.png',
'output_4.png', str(r), str(option)],
stdout=None, stderr=None)
self.wait_proc([p1, p2, p3, p4], timeout)
#Compute histograms of images
im1 = image(self.work_dir + 'input_0.sel.png')
im2 = image(self.work_dir + 'output_1.png')
im3 = image(self.work_dir + 'output_2.png')
im4 = image(self.work_dir + 'output_3.png')
im5 = image(self.work_dir + 'output_4.png')
#compute maximum of histogram values for the input image
maxH = im1.max_histogram(option="all")
#draw all the histograms using the same reference maximum
im1.histogram(option="all", maxRef=maxH)
im1.save(self.work_dir + 'input_0_hist.png')
im2.histogram(option="all", maxRef=maxH)
im2.save(self.work_dir + 'output_1_hist.png')
im3.histogram(option="all", maxRef=maxH)
im3.save(self.work_dir + 'output_2_hist.png')
im4.histogram(option="all", maxRef=maxH)
im4.save(self.work_dir + 'output_3_hist.png')
im5.histogram(option="all", maxRef=maxH)
im5.save(self.work_dir + 'output_4_hist.png')
def result(self):
"""
display the algo results
"""
# read the parameters
L = self.cfg['param']['l']
s = self.cfg['param']['s']
sizeX = image(self.work_dir + 'input_0.png').size[0]
sizeY = max(image(self.work_dir + 'input_0.png').size[1],
3 * 128 + 4 * 20)
return self.tmpl_out("result.html", L=L, s=s,
sizeX=sizeX, sizeY=sizeY)
def grid(self, stepG, scaleR, action=None, x=0, y=0):
"""
handle the grid drawing and selection
"""
if action == 'run':
# use the whole image
img = image(self.work_dir + 'input_0.png')
img.save(self.work_dir + 'input' + self.input_ext)
img.save(self.work_dir + 'input.png')
# go to the wait page, with the key and scale
stepG = 0
http.redir_303(self.base_url
+ "wait?key=%s&scaleR=%s&step=%s"
% (self.key, scaleR, stepG))
return
elif action == 'redraw':
# draw the grid
step = int(stepG)
if 0 < step:
img = image(self.work_dir + 'input_0.png')
img.draw_grid(step)
img.save(self.work_dir + 'input_grid.png')
grid = True
else:
grid = False
return self.tmpl_out("params.html", step=stepG,
grid=grid, scale_r=float(scaleR))
else:
# use a part of the image
x = int(x)
y = int(y)
# get the step used to draw the grid
step = int(stepG)
assert step > 0
# cut the image section
img = image(self.work_dir + 'input_0.png')
x0 = (x / step) * step
y0 = (y / step) * step
x1 = min(img.size[0], x0 + step)
y1 = min(img.size[1], y0 + step)
img.crop((x0, y0, x1, y1))
# zoom and save image
img.resize((400, 400), method="nearest")
img.save(self.work_dir + 'input' + self.input_ext)
img.save(self.work_dir + 'input.png')
# go to the wait page, with the key and scale
http.redir_303(self.base_url
+ "wait?key=%s&scaleR=%s&step=%s"
% (self.key, scaleR, stepG))
return
def result(self):
"""
display the algo results
"""
# read the parameters
s1 = self.cfg['param']['s1']
s2 = self.cfg['param']['s2']
sizeY = image(self.work_dir + 'input_0.png').size[1]
sizeYhist = image(self.work_dir + 'input_0_hist.png').size[1]
# add 20 pixels to the histogram size to take margin into account
sizeYmax = max(sizeY, sizeYhist+60)
return self.tmpl_out("result.html", s1=s1, s2=s2,
sizeY="%i" % sizeYmax)
def result(self):
"""
display the algo results
"""
return self.tmpl_out("result.html",
useOriginal=os.path.isfile(self.work_dir + 'input_0s.png'),
sizeY="%i" % image(self.work_dir + 'input.png').size[1])
def wait(self, **kwargs):
"""
params handling and run redirection
"""
if 'sense' in kwargs:
self.cfg['param']['sense'] = kwargs['sense']
self.cfg.save()
if 'norectif' in kwargs:
self.cfg['param']['norectif'] = True
self.cfg.save()
try:
if 'dmin' in kwargs: # Check integer values
dmin = int(kwargs['dmin'])
dmax = int(kwargs['dmax'])
self.cfg['param']['dmin'] = dmin
self.cfg['param']['dmax'] = dmax
self.cfg.save()
except ValueError:
return self.error(
errcode='badparams',
errmsg="The disparity range parameters must be integers.")
http.refresh(self.base_url + 'run?key=%s' % self.key)
return self.tmpl_out("wait.html",
height=image(self.work_dir +
'input_0.png').size[1])
def result(self, public=None):
"""
display the algo results
"""
return self.tmpl_out("result.html",
height=image(self.work_dir
+ 'input_0.png').size[1])
def input_select(self, **kwargs):
"""
use the selected available input images
"""
print("ENTERING input_select")
self.new_key()
self.init_cfg()
print("key = " + self.key)
# kwargs contains input_id.x and input_id.y
input_id = kwargs.keys()[0].split('.')[0]
assert input_id == kwargs.keys()[1].split('.')[0]
# get the images
input_dict = config.file_dict(self.input_dir)
idir = self.input_dir + input_dict[input_id]['subdir'] + '/'
fnames = "a.png b.png".split()
for f in fnames:
shutil.copy(idir + f, self.work_dir + f)
hastruth = os.path.isfile(idir + "t.tiff")
self.cfg['meta']['hastruth'] = hastruth
if (hastruth):
shutil.copy(idir + "t.tiff", self.work_dir + "t.tiff")
shutil.copy(idir + "t.png", self.work_dir + "t.png")
#fnames = input_dict[input_id]['files'].split()
#for i in range(len(fnames)):
# shutil.copy(self.input_dir + fnames[i],
# self.work_dir + 'input_%i' % i)
#msg = self.process_input()
self.log("input selected : %s" % input_id)
self.cfg['meta']['original'] = False
self.cfg['meta']['height'] = image(self.work_dir + '/a.png').size[1]
self.cfg.save()
# jump to the params page
return self.params(msg="(no message)", key=self.key)
def set_default_gui_state(self):
img = image(self.work_dir + "input_0.png")
w = img.size[0]
h = img.size[1]
m = 50
ow = w + 2 * m
oh = h + 2 * m
self.cfg["param"]["m"] = m
self.cfg["param"]["w"] = w
self.cfg["param"]["h"] = h
self.cfg["param"]["ow"] = ow
self.cfg["param"]["oh"] = oh
self.cfg["param"]["c"] = str(
[
[(0 + m + int(w * 0.4)), 0 + m + int(h * 0.4)],
[w - 1 + m, 0 + m],
[w - 1 + m, h - 1 + m],
[0 + m, h - 1 + m],
]
)
self.cfg["param"]["p"] = str([[0, 0], [w - 1, 0], [w - 1, h - 1], [0, h - 1]])
self.cfg["param"]["hit"] = -1
self.cfg["param"]["state"] = "plain"
return
def index(self):
"""
demo presentation and input menu
"""
#print("ENTERING index")
tn_size = 192
# read the input index as a dict
inputd = config.file_dict(self.input_dir)
#print(inputd)
for (input_id, input_info) in inputd.items():
tn_fname = [thumbnail(self.input_dir + \
input_info['subdir'] \
+ '/i0000.png',(tn_size,tn_size))]
inputd[input_id]['hastruth'] = os.path.isfile(
self.input_dir + input_info['subdir'] + '/a.png')
inputd[input_id]['height'] = image(self.input_dir +
input_info['subdir'] +
'/i0000.png').size[1]
inputd[input_id]['baseinput'] = \
self.input_url + input_info['subdir'] + '/'
inputd[input_id]['url'] = [
self.input_url + input_info['subdir'] + '/' +
os.path.basename(f) for f in ["i0000.png"]
]
inputd[input_id]['tn_url'] = [
self.input_url + input_info['subdir'] + '/' +
os.path.basename(f) for f in tn_fname
]
return self.tmpl_out("input.html", inputd=inputd)
def input_select(self, **kwargs):
"""
use the selected available input images
"""
#print("ENTERING input_select")
self.new_key()
self.init_cfg()
#print("key = " + self.key)
# kwargs contains input_id.x and input_id.y
input_id = kwargs.keys()[0].split('.')[0]
assert input_id == kwargs.keys()[1].split('.')[0]
# get the images
input_dict = config.file_dict(self.input_dir)
idir = self.input_dir + input_dict[input_id]['subdir'] + '/'
#print("idir = " + idir)
fnames = [os.path.basename(f) for f in glob.glob(idir + 'i????.png')]
for f in fnames:
shutil.copy(idir + f, self.work_dir + f)
hastruth = os.path.isfile(idir + "a.png")
self.cfg['meta']['hastruth'] = hastruth
self.cfg['meta']['maxframes'] = len(fnames)
if (hastruth):
shutil.copy(idir + "a.png", self.work_dir + "a.png")
self.log("input selected : %s" % input_id)
self.cfg['meta']['original'] = False
self.cfg['meta']['height'] = image(self.work_dir +
'/i0000.png').size[1]
self.cfg.save()
# jump to the params page
return self.params(msg="(no message)", key=self.key)
def result(self):
"""
display the algo results
"""
# read the parameters
s1 = self.cfg['param']['s1']
s2 = self.cfg['param']['s2']
sizeY = image(self.work_dir + 'input_0.png').size[1]
sizeYhist = image(self.work_dir + 'input_0_hist.png').size[1]
# add 20 pixels to the histogram size to take margin into account
sizeYmax = max(sizeY, sizeYhist + 60)
return self.tmpl_out("result.html",
s1=s1,
s2=s2,
sizeY="%i" % sizeYmax)
def result(self):
"""
display the algo results
"""
print "Display results"
# read the parameters
sigma = self.cfg['param']['sigma']
try:
x0 = self.cfg['param']['x0']
except KeyError:
x0 = None
try:
y0 = self.cfg['param']['y0']
except KeyError:
y0 = None
try:
x1 = self.cfg['param']['x1']
except KeyError:
x1 = None
try:
y1 = self.cfg['param']['y1']
except KeyError:
y1 = None
(sizeX, sizeY) = image(self.work_dir + 'input_0.sel.png').size
zoom_factor = None
return self.tmpl_out("result.html", sigma=sigma, \
x0=x0, y0=y0, x1=x1, y1=y1, \
sizeX=sizeX, sizeY=sizeY, \
zoom_factor=zoom_factor)
def result(self, public=None):
"""
display the algo results
"""
return self.tmpl_out("result.html",
height=image(self.work_dir +
'input_0.png').size[1])
def result(self):
"""
display the algo results
"""
sizeY = image(self.work_dir + 'input_0.png').size[1]
return self.tmpl_out("result.html", sizeY=sizeY)
def run_algo(self):
"""Core algorithm runner, it could also be called by a batch processor,
this one needs no parameter.
"""
# run LSD
p = self.run_proc(['lsd', '-P', 'output.eps', '-S', 'output.svg',
'input_0_selection.pgm', 'output.txt'])
self.wait_proc(p)
# convert the EPS result into a PNG image
try:
p = self.run_proc(['/usr/bin/gs', '-dNOPAUSE', '-dBATCH',
'-sDEVICE=pnggray', '-dGraphicsAlphaBits=4',
'-r72','-dEPSCrop', '-sOutputFile=output.png',
'output.eps'])
self.wait_proc(p)
im = image(self.work_dir + "output.png")
im.convert('1x8i')
im.invert()
im.save(self.work_dir + "output-inv.png")
except Exception:
self.log("eps->png conversion failed,"
+ " GS is probably missing on this system")
return
def input_select(self, **kwargs):
"""
use the selected available input images
"""
print("ENTERING input_select")
self.new_key()
self.init_cfg()
print("key = " + self.key)
# kwargs contains input_id.x and input_id.y
input_id = kwargs.keys()[0].split('.')[0]
assert input_id == kwargs.keys()[1].split('.')[0]
# get the images
input_dict = config.file_dict(self.input_dir)
idir = self.input_dir + input_dict[input_id]['subdir'] + '/'
print("idir = " + idir)
fnames = [os.path.basename(f) for f in glob.glob(idir + 'i????.png')]
for f in fnames:
shutil.copy(idir + f, self.work_dir + f)
hastruth = os.path.isfile(idir + "a.png")
self.cfg['meta']['hastruth'] = hastruth
self.cfg['meta']['maxframes'] = len(fnames)
if (hastruth):
shutil.copy(idir + "a.png", self.work_dir + "a.png")
self.log("input selected : %s" % input_id)
self.cfg['meta']['original'] = False
self.cfg['meta']['height'] = image(self.work_dir + '/i0000.png').size[1]
self.cfg.save()
# jump to the params page
return self.params(msg="(no message)", key=self.key)
def run_algo(self, stdout=None):
"""
The core algo runner
could also be called by a batch processor
this one needs no parameter
"""
timeout = False
omega_string = str(self.cfg['param']['omega'])
if omega_string == 'G':
omega_string += ':' + str(self.cfg['param']['gaussianstd'])
if self.cfg['param']['method'] == 'poly':
method_string = 'poly:' + str(self.cfg['param']['degree'])
else:
method_string = 'interp:' + str(self.cfg['param']['numlevels'])
self.wait_proc([
self.run_proc([
'ace', '-a',
str(self.cfg['param']['alpha']), '-w' + omega_string,
'-m' + method_string, 'input_0_sel.png', 'ace.png'
],
stdout=stdout,
stderr=stdout),
self.run_proc(['histeq', 'input_0_sel.png', 'he.png'],
stdout=None,
stderr=stdout)
], timeout)
self.cfg['param']['displayheight'] = max(200, \
image(self.work_dir + 'input_0_sel.png').size[1])
self.cfg.save()
def result(self):
"""
display the algo results
"""
self.cfg['meta']['height'] = image(self.work_dir +
'/i0000.png').size[1]
return self.tmpl_out("result.html")
def wait(self, **kwargs):
"""
params handling and run redirection
"""
if 'rectif' in kwargs:
self.cfg['param']['rectif'] = True
if 'valK' in kwargs:
self.autoK = False
self.cfg['param']['k'] = \
str(min(self.Kmax, float(kwargs['valK'])))
if 'autoK' in kwargs and kwargs['autoK'] == 'True':
self.autoK = True
self.cfg['param']['k'] = 'auto'
if 'valL' in kwargs:
self.autoL = False
self.cfg['param']['lambda'] = \
str(min(self.Lmax, float(kwargs['valL'])))
if 'autoL' in kwargs and kwargs['autoL'] == 'True':
self.autoL = True
self.cfg['param']['lambda'] = 'auto'
if 'dmin' in kwargs: # Check integer values
dmin = int(kwargs['dmin'])
dmax = int(kwargs['dmax'])
self.cfg['param']['dmin'] = dmin
self.cfg['param']['dmax'] = dmax
self.cfg.save()
http.refresh(self.base_url + 'run?key=%s' % self.key)
return self.tmpl_out("wait.html",
height=image(self.work_dir
+ 'input_0.png').size[1])
def result(self):
"""
display the algo results
"""
print "Display results"
# read the parameters
sigma = self.cfg['param']['sigma']
try:
x0 = self.cfg['param']['x0']
except KeyError:
x0 = None
try:
y0 = self.cfg['param']['y0']
except KeyError:
y0 = None
try:
x1 = self.cfg['param']['x1']
except KeyError:
x1 = None
try:
y1 = self.cfg['param']['y1']
except KeyError:
y1 = None
(sizeX, sizeY) = image(self.work_dir + 'input_0.sel.png').size
zoom_factor = None
return self.tmpl_out("result.html", sigma=sigma, \
x0=x0, y0=y0, x1=x1, y1=y1, \
sizeX=sizeX, sizeY=sizeY, \
zoom_factor=zoom_factor)
def index(self):
"""
demo presentation and input menu
"""
print("ENTERING index")
tn_size = 192
# read the input index as a dict
inputd = config.file_dict(self.input_dir)
print(inputd)
for (input_id, input_info) in inputd.items():
fname = ["a.png", "b.png", "t.png"]
tn_fname = [thumbnail(self.input_dir + \
input_info['subdir'] \
+ '/a.png',(tn_size,tn_size))]
inputd[input_id]['hastruth'] = os.path.isfile(
self.input_dir + input_info['subdir']+'/t.tiff')
inputd[input_id]['height'] = image(self.input_dir
+ input_info['subdir'] + '/a.png').size[1]
inputd[input_id]['baseinput'] = \
self.input_url + input_info['subdir'] + '/'
inputd[input_id]['url'] = [self.input_url
+ input_info['subdir']
+ '/' + os.path.basename(f)
for f in fname]
inputd[input_id]['tn_url'] = [self.input_url
+ input_info['subdir']
+ '/'
+ os.path.basename(f)
for f in tn_fname]
return self.tmpl_out("input.html", inputd=inputd)
def result(self):
"""
display the algo results
"""
sizeY=image(self.work_dir + 'input_0.png').size[1]
return self.tmpl_out("result.html", sizeY=sizeY)
def input_select(self, **kwargs):
"""
use the selected available input images
"""
print("ENTERING input_select")
self.new_key()
self.init_cfg()
print("key = " + self.key)
# kwargs contains input_id.x and input_id.y
input_id = kwargs.keys()[0].split('.')[0]
assert input_id == kwargs.keys()[1].split('.')[0]
# get the images
input_dict = config.file_dict(self.input_dir)
idir = self.input_dir + input_dict[input_id]['subdir'] + '/'
fnames = "a.png b.png".split()
for f in fnames:
shutil.copy(idir + f, self.work_dir + f)
hastruth = os.path.isfile(idir + "t.tiff")
self.cfg['meta']['hastruth'] = hastruth
if (hastruth):
shutil.copy(idir + "t.tiff", self.work_dir + "t.tiff")
shutil.copy(idir + "t.png", self.work_dir + "t.png")
#fnames = input_dict[input_id]['files'].split()
#for i in range(len(fnames)):
# shutil.copy(self.input_dir + fnames[i],
# self.work_dir + 'input_%i' % i)
#msg = self.process_input()
self.log("input selected : %s" % input_id)
self.cfg['meta']['original'] = False
self.cfg['meta']['height'] = image(self.work_dir + '/a.png').size[1]
self.cfg.save()
# jump to the params page
return self.params(msg="(no message)", key=self.key)
def _rectify(self, stdout):
"""
rectify input images
"""
start_time = time.time()
end_time = start_time + self.timeout/3
(w, h) = image(self.work_dir+'input_0.png').size
pairs = self.work_dir+'input_0.png_input_1.png_pairs.txt'
pairs_good = self.work_dir+'input_0.png_input_1.png_pairs_orsa.txt'
try:
p = self.run_proc(['sift',
self.work_dir+'input_0.png',
self.work_dir+'input_1.png',
pairs],
stdout=stdout, stderr=stdout)
self.wait_proc(p, timeout=end_time-start_time)
start_time = time.time()
p = self.run_proc(['orsa', str(w), str(h),
pairs, pairs_good, '500', '1', '0', '2', '0'],
stdout=stdout, stderr=stdout)
self.wait_proc(p, timeout=end_time-start_time)
start_time = time.time()
outRect = open(self.work_dir+'rect.txt','w')
p = self.run_proc(['rectify', pairs_good, str(w), str(h),
self.work_dir+'input_0.png_h.txt',
self.work_dir+'input_1.png_h.txt'],
stdout=outRect, stderr=outRect)
self.wait_proc(p, timeout=end_time-start_time)
start_time = time.time()
p = []
p.append(self.run_proc(['homography', '-f',
self.work_dir+'input_0.png',
self.work_dir+'input_0.png_h.txt',
self.work_dir+'H_input_0.png'],
stdout=outRect, stderr=outRect))
p.append(self.run_proc(['homography', '-f',
self.work_dir+'input_1.png',
self.work_dir+'input_1.png_h.txt',
self.work_dir+'H_input_1.png'],
stdout=outRect, stderr=outRect))
self.wait_proc(p, timeout=end_time-start_time)
except RuntimeError:
if 0 != p.returncode:
stdout.close()
raise NoMatchError
else:
raise
outRect.close()
outRect = open(self.work_dir+'rect.txt','r')
lines = outRect.readlines()
for line in lines:
stdout.write(line)
words = line.split()
if words[0] == "Disparity:":
dmin = words[1]
dmax = words[2]
outRect.close()
os.unlink(self.work_dir+'rect.txt')
stdout.flush()
return (dmin, dmax)
def run_algo(self):
"""Core algorithm runner, it could also be called by a batch processor,
this one needs no parameter.
"""
# run LSD
p = self.run_proc([
'lsd', '-P', 'output.eps', '-S', 'output.svg',
'input_0_selection.pgm', 'output.txt'
])
self.wait_proc(p)
# convert the EPS result into a PNG image
try:
p = self.run_proc([
'/usr/bin/gs', '-dNOPAUSE', '-dBATCH', '-sDEVICE=pnggray',
'-dGraphicsAlphaBits=4', '-r72', '-dEPSCrop',
'-sOutputFile=output.png', 'output.eps'
])
self.wait_proc(p)
im = image(self.work_dir + "output.png")
im.convert('1x8i')
im.invert()
im.save(self.work_dir + "output-inv.png")
except Exception:
self.log("eps->png conversion failed," +
" GS is probably missing on this system")
return
def scan_for_faces(img):
print("Scanning " + img + " for faces.")
global df_faces
try:
time.sleep(3)
print("Creating image")
this_image = image(img)
print("Getting faces")
faces = this_image.get_image_faces()
print("Image: " + img)
print("Number of faces: " + str(len(faces)))
df_faces = df_faces.append(pd.DataFrame({
'img':img,
'left':[f.left for f in faces],
'top':[f.top for f in faces],
'width':[f.width for f in faces],
'height':[f.height for f in faces],
'age':[f.age for f in faces],
'gender':[f.gender for f in faces],
'smile':[f.smile for f in faces],
'emotion':[f.emotion for f in faces]
}))
return True
except:
return False
def result(self):
"""
display the algo results
"""
return self.tmpl_out("result.html",
height=image(self.work_dir
+ 'output.png').size[1])
def run_algo(self, params=None):
"""
the core algo runner
could also be called by a batch processor
this one needs no parameter
"""
self.cfg['param']['sizex'] = image(self.work_dir + \
'input_0.png').size[0]
self.cfg['param']['sizey'] = image(self.work_dir + \
'input_0.png').size[1]
commandargs = ['LUTBasedNSDistanceTransform']
print(type(self.cfg['param']['distance_def']), \
self.cfg['param']['distance_def'])
if self.cfg['param']['centered']:
commandargs += ['-c']
if self.cfg['param']['distance_def'] == 'd4':
commandargs += ['-4']
elif self.cfg['param']['distance_def'] == 'd8':
commandargs += ['-8']
elif self.cfg['param']['distance_def'] == 'ratio':
commandargs += ['-r', str(self.cfg['param']['ratio'])]
elif self.cfg['param']['distance_def'] == 'sequence':
commandargs += ['-s', str(self.cfg['param']['sequence'])]
commandargs += ['-f', 'input_0.png']
commandargs += ['-t', 'png']
f = open(self.work_dir+"resu_r.png", "w")
self.runCommand(commandargs, stdOut=f, stdErr=subprocess.PIPE, \
comp='> resu_r.png')
f.close()
commandargs = ['convert.sh', '-normalize', 'resu_r.png', \
'resu_n.jpg']
self.runCommand(commandargs, stdErr=subprocess.PIPE)
fcommands = open(self.work_dir+"commands.txt", "w")
fcommands.write(self.commands)
fcommands.close()
return
def result(self, public=None):
"""
display the algo results
SHOULD be defined in the derived classes, to check the parameters
"""
return self.tmpl_out('result.html',
displaysize=image(self.work_dir + 'output_0.bmp').size,
stdout=open(self.work_dir + 'stdout.txt', 'r').read())
def result(self):
"""
display the algo results
"""
self.cfg['param']['orsa']= '0'
height = 10 + image(self.work_dir+'matches.png').size[1]
return self.tmpl_out("result.html", height=height)
def plot_cross(input_image, x, y, output_image):
"""
draw a cross at the position (x,y)
"""
img = image(input_image)
img.draw_cross((x, y), size=4, color="white")
img.draw_cross((x, y), size=2, color="red")
img.save(output_image)
def run_algo(self, params=None):
"""
the core algo runner
could also be called by a batch processor
this one needs no parameter
"""
self.cfg['param']['sizex'] = image(self.work_dir + \
'input_0.png').size[0]
self.cfg['param']['sizey'] = image(self.work_dir + \
'input_0.png').size[1]
commandargs = ['LUTBasedNSDistanceTransform']
print(type(self.cfg['param']['distance_def']), \
self.cfg['param']['distance_def'])
if self.cfg['param']['centered']:
commandargs += ['-c']
if self.cfg['param']['distance_def'] == 'd4':
commandargs += ['-4']
elif self.cfg['param']['distance_def'] == 'd8':
commandargs += ['-8']
elif self.cfg['param']['distance_def'] == 'ratio':
commandargs += ['-r', str(self.cfg['param']['ratio'])]
elif self.cfg['param']['distance_def'] == 'sequence':
commandargs += ['-s', str(self.cfg['param']['sequence'])]
commandargs += ['-f', 'input_0.png']
commandargs += ['-t', 'png']
f = open(self.work_dir + "resu_r.png", "w")
self.runCommand(commandargs, stdOut=f, stdErr=subprocess.PIPE, \
comp='> resu_r.png')
f.close()
commandargs = ['convert.sh', '-normalize', 'resu_r.png', \
'resu_n.jpg']
self.runCommand(commandargs, stdErr=subprocess.PIPE)
fcommands = open(self.work_dir + "commands.txt", "w")
fcommands.write(self.commands)
fcommands.close()
return
def result(self):
"""
display the algo results
"""
# read the parameters
t = self.cfg['param']['t']
return self.tmpl_out("result.html", t=t,
sizeY="%i" % image(self.work_dir
+ 'input_1.png').size[1])