def test_set_color_with_alpha():
img = np.zeros((10, 10))
rr, cc, alpha = line_aa(0, 0, 0, 30)
set_color(img, (rr, cc), 1, alpha=alpha)
# Wrong dimensionality color
assert_raises(ValueError, set_color, img, (rr, cc), (255, 0, 0), alpha=alpha)
img = np.zeros((10, 10, 3))
rr, cc, alpha = line_aa(0, 0, 0, 30)
set_color(img, (rr, cc), (1, 0, 0), alpha=alpha)
def test_line_equal_aliasing_horizontally_vertically():
img0 = np.zeros((25, 25))
img1 = np.zeros((25, 25))
# Near-horizontal line
rr, cc, val = line_aa(10, 2, 12, 20)
img0[rr, cc] = val
# Near-vertical (transpose of prior)
rr, cc, val = line_aa(2, 10, 20, 12)
img1[rr, cc] = val
# Difference - should be zero
assert_array_equal(img0, img1.T)
def test_line_aa_horizontal():
img = np.zeros((10, 10))
rr, cc, val = line_aa(0, 0, 0, 9)
set_color(img, (rr, cc), 1, alpha=val)
img_ = np.zeros((10, 10))
img_[0, :] = 1
assert_array_equal(img, img_)
def test_line_aa_horizontal():
img = np.zeros((10, 10))
rr, cc, val = line_aa(0, 0, 0, 9)
img[rr, cc] = val
img_ = np.zeros((10, 10))
img_[0, :] = 1
assert_array_equal(img, img_)
def test_line_aa_vertical():
img = np.zeros((10, 10))
rr, cc, val = line_aa(0, 0, 9, 0)
img[rr, cc] = val
img_ = np.zeros((10, 10))
img_[:, 0] = 1
assert_array_equal(img, img_)
def crop_to_sequence(self, track_list, currSharp):
'''
In the crop filename, I add the number of the image in the galerie so that they're by default ordered in time
I need the bounding box of the membrane and also to use whitefield images.
'''
for track in track_list:
id_=track.id
lstFrames=sorted(track.lstPoints.keys(), key=itemgetter(0))
rr=[]; cc=[]; val=[]; nextCoord=None
for i, el in enumerate(lstFrames):
im, cell_id=el
coordonnees=track.lstPoints[(im, cell_id)] if nextCoord==None else nextCoord
try:
nextCoord=track.lstPoints[lstFrames[i+1]]
except IndexError:
continue
else:
r,c,v=draw.line_aa(coordonnees[0], coordonnees[1],nextCoord[0], nextCoord[1])
rr.extend(r); cc.extend(c); val.extend(v)
for im, cell_id in lstFrames:
#renumbering according to xb screen/PCNA image numbering
local_im=im+1
#draw a dot on the cell which is followed
cell_x, cell_y=track.lstPoints[(im, cell_id)]
dot_rr, dot_cc=draw.circle(cell_x, cell_y, radius=3)
image_name= self.settings.imageFilename.format(self.well, local_im)
image=vi.readImage(os.path.join(self.settings.allDataFolder, self.plate, 'analyzed', self.well, 'images/tertiary_contours_expanded', image_name))
#X,x,x__, Y,y,y__=self._newImageSize(crop_coordinates)
x__=self.settings.XMAX; y__=self.settings.YMAX; x=0; y=0; X=self.settings.XMAX; Y=self.settings.YMAX
croppedImage = VigraArray((x__, y__, 3), dtype=np.dtype('float32'))
croppedImage=image[x:X, y:Y]
croppedImage[rr,cc,0]=np.array(val)*255
#If there is a sharp movement, the cell center is pinky red
if im in currSharp[id_]:
croppedImage[dot_rr, dot_cc, 0]=242
croppedImage[dot_rr, dot_cc, 1]=21
croppedImage[dot_rr, dot_cc, 2]=58
else:
#If not, it is green
croppedImage[dot_rr, dot_cc, 1]=255
vi.writeImage(croppedImage, \
os.path.join(self.outputFolder, self.plate, 'galerie',
self.settings.outputImage.format(self.plate, self.well.split('_')[0],id_, im)),\
dtype=np.dtype('uint8'))
return
def test_line_aa_diagonal():
img = np.zeros((10, 10))
rr, cc, val = line_aa(0, 0, 9, 6)
img[rr, cc] = 1
# Check that each pixel belonging to line,
# also belongs to line_aa
r, c = line(0, 0, 9, 6)
for x, y in zip(r, c):
assert_equal(img[r, c], 1)
def add_boundary(self, boundaries):
"""
Add boundaries to the map.
:param boundaries: list of tuples containing coordinates of boundaries'
start and end points
"""
# Extend list of boundaries
self.boundaries.extend(boundaries)
# Iterate over list of boundaries
for boundary in boundaries:
sx = self.w2m(boundary[0][0], boundary[0][1])
gx = self.w2m(boundary[1][0], boundary[1][1])
path_x, path_y, _ = line_aa(sx[0], sx[1], gx[0], gx[1])
for x, y in zip(path_x, path_y):
self.data[y, x] = 0
def __draw_line(self, new_c, new_r):
r0 = int(round(self.__clip_coordinate(self.__r, 0)))
c0 = int(round(self.__clip_coordinate(self.__c, 1)))
r1 = int(round(self.__clip_coordinate(new_r, 0)))
c1 = int(round(self.__clip_coordinate(new_c, 1)))
if self.aa:
rr, cc, val = line_aa(r0, c0, r1, c1)
else:
rr, cc = line(r0, c0, r1, c1)
val = 1
if self.__channels == 1:
self.array[rr, cc] = val * self.__color
else:
for c in range(self.__channels):
self.array[rr, cc, c] = val * self.__color[c]
def _get_min_width(self, wp, t_x, t_y, max_width):
"""
Compute the minimum distance between the current waypoint and the
orthogonal cell on the border of the path
:param wp: current waypoint
:param t_x: x coordinate of border cell in map coordinates
:param t_y: y coordinate of border cell in map coordinates
:param max_width: maximum path width in m
:return: min_width to border and corresponding cell
"""
# Get neighboring cells of orthogonal cell (account for
# discretization inaccuracy)
tn_x, tn_y = [], []
for i in range(-1, 2, 1):
for j in range(-1, 2, 1):
tn_x.append(t_x + i)
tn_y.append(t_y + j)
# Get pixel coordinates of waypoint
wp_x, wp_y = self.map.w2m(wp.x, wp.y)
# Get Bresenham paths to all possible cells
paths = []
for t_x, t_y in zip(tn_x, tn_y):
x_list, y_list, _ = line_aa(wp_x, wp_y, t_x, t_y)
paths.append(zip(x_list, y_list))
# Compute minimum distance to border cell
min_width = max_width
# map inspected cell to world coordinates
min_cell = self.map.m2w(t_x, t_y)
for path in paths:
for cell in path:
t_x, t_y = cell[0], cell[1]
# If path goes through occupied cell
if self.map.data[t_y, t_x] == 0:
# Get world coordinates
c_x, c_y = self.map.m2w(t_x, t_y)
cell_dist = np.sqrt((wp.x - c_x)**2 + (wp.y - c_y)**2)
if cell_dist < min_width:
min_width = cell_dist
min_cell = (c_x, c_y)
return min_width, min_cell
def _render_line(canvas, curr_pos, l_args, absolute, color):
"""Renders a line in the given canvas."""
end_point = l_args
if not absolute:
end_point[0] += curr_pos[0]
end_point[1] += curr_pos[1]
rr, cc, val = draw.line_aa(int(curr_pos[0]), int(curr_pos[1]),
int(end_point[0]), int(end_point[1]))
max_possible = len(canvas)
within_range = lambda x: 0 <= x < max_possible
filtered = [(x, y, v) for x, y, v in zip(rr, cc, val)
if within_range(x) and within_range(y)]
if not filtered:
return
rr, cc, val = list(zip(*filtered))
val = [(v * color) for v in val]
canvas[cc, rr, :] = val
def make_lines(nbins=9, degree_base=180, frame_size=8):
"""
#TODO docstring
"""
# make nbins channel image
lines_bank = np.zeros((nbins, frame_size, frame_size), dtype=np.uint8)
# angles by bins, convert to radians
theta_rad = (np.arange(nbins) * (degree_base / nbins)) * (np.pi / 180)
# line diameter
_d = frame_size - 1
# find m and c (y=mx+c) passing center point of visualization image (for now, using square (w=h))
m = np.tan(theta_rad)
c = (_d + 1 // 2) * (1 - m)
# find max min y,x
x_1 = np.abs(1 + (_d) // 2 * (1 + np.cos(theta_rad))).astype(np.int)
y_1 = np.abs(1 + (_d) // 2 * (1 + np.sin(theta_rad))).astype(np.int)
y_0 = _d - y_1
x_0 = _d - x_1
for i, img in enumerate(lines_bank):
# using skimage
# rr, cc = draw.line(x_1[i], y_0[i], x_0[i], y_1[i])
# # x is y, y is -x (rotate by 90deg)
# # rr, cc = draw.line(y_0[i], -x_0[i], y_1[i], -x_1[i])
# img[rr,cc] = 255
rr, cc, val = draw.line_aa(x_1[i], y_0[i], x_0[i], y_1[i])
img[rr, cc] = val * 255
# # using PIL
# pilimg = Image.new('L', (frame_size, frame_size), 0)
# draw = ImageDraw.Draw(pilimg)
# # draw.line([(x_1[i], y_0[i]), (x_0[i], y_1[i])], fill=255, width=2)
# draw.line([(y_1[i], x_0[i]), (y_0[i], x_1[i])], fill=128, width=1)
# # print(np.asarray(pilimg).shape,img.shape)
# img += np.asarray(pilimg)
#DEBUG
# plt.imshow(img)
# plt.show()
return lines_bank
def projection_on_magnet(ends_of_strawtubes, x_resolution=1., y_resolution=1., z_magnet=3070.):
tubes = []
for i in range(len(ends_of_strawtubes)):
tubes.append(ends2params(ends_of_strawtubes[i]))
lines = []
z3 = z_magnet
for i in range(len(tubes)):
for j in range(i+1, len(tubes)):
t1 = tubes[i]
t2 = tubes[j]
if (t1[2] != t2[2]) and (np.sum(t1[1] - t2[1]) < 0.01):
start1 = t1[0]
start2 = t2[0]
z1 = t1[2]
z2 = t2[2]
start3 = (start2 - start1) / (z2 - z1) * (z3 - z2) + start2
if (start3[1] > -500) and (start3[1] < 500) and (start3[0] < -225) and (start3[0] > -275):
lines.append([start3, t1[1]])
x_len = int(600 * x_resolution)
y_len = int(1200 * y_resolution)
line_len = 500
matrix = np.zeros([x_len, y_len])
for line in lines:
rr, cc, val = line_aa(int(round(line[0][0] * x_resolution)) + x_len / 2, int(round(line[0][1] * y_resolution)) + y_len / 2,\
int(round((line[0][0] + line[1][0] * line_len) * x_resolution)) + x_len / 2,\
int(round((line[0][1] + line[1][1] * line_len) * y_resolution)) + y_len / 2)
matrix[rr, cc] += 1
return matrix
def draw_joint(colors, pose_joints, joint_line_list, radius=2):
im_size = (colors.shape[0], colors.shape[1])
for f, t in joint_line_list:
from_missing = pose_joints[0,f] == MISSING_VALUE or pose_joints[1,f] == MISSING_VALUE
to_missing = pose_joints[0,t] == MISSING_VALUE or pose_joints[1,t] == MISSING_VALUE
if from_missing or to_missing:
continue
yy, xx, val = line_aa(pose_joints[0,f], pose_joints[1,f], pose_joints[0,t], pose_joints[1,t])
yy, xx = np.clip(yy, 0, im_size[0]-1), np.clip(xx, 0, im_size[1] - 1)
colors[yy, xx] = np.expand_dims(val, 1) * 255
# mask[yy, xx] = True
colormap = labelcolormap(pose_joints.shape[1])
for i in range(pose_joints.shape[1]):
if pose_joints[0,i] == MISSING_VALUE or pose_joints[1,i] == MISSING_VALUE:
continue
yy, xx = disk((pose_joints[0,i], pose_joints[1,i]), radius=radius, shape=im_size)
colors[yy, xx] = colormap[i]
return colors
def projection_on_magnet(ends_of_strawtubes, x_resolution=1., y_resolution=1., z_magnet=3070.):
tubes = []
for i in range(len(ends_of_strawtubes)):
tubes.append(ends2params(ends_of_strawtubes[i]))
lines = []
z3 = z_magnet
for i in range(len(tubes)):
for j in range(i+1, len(tubes)):
t1 = tubes[i]
t2 = tubes[j]
if (t1[2] != t2[2]) and (np.sum(t1[1] - t2[1]) < 0.01):
start1 = t1[0]
start2 = t2[0]
z1 = t1[2]
z2 = t2[2]
start3 = (start2 - start1) / (z2 - z1) * (z3 - z2) + start2
if (start3[1] > -500) and (start3[1] < 500) and (start3[0] < -225) and (start3[0] > -275):
lines.append([start3, t1[1]])
x_len = int(600 * x_resolution)
y_len = int(1200 * y_resolution)
line_len = 500
matrix = np.zeros([x_len, y_len])
for line in lines:
rr, cc, val = line_aa(int(round(line[0][0] * x_resolution)) + x_len / 2, int(round(line[0][1] * y_resolution)) + y_len / 2,\
int(round((line[0][0] + line[1][0] * line_len) * x_resolution)) + x_len / 2,\
int(round((line[0][1] + line[1][1] * line_len) * y_resolution)) + y_len / 2)
matrix[rr, cc] += 1
return matrix
def get_line():
def sample():
r1 = int(rng.random() * image.shape[0])
c1 = int(rng.random() * image.shape[1])
r2 = int(rng.random() * image.shape[0])
c2 = int(rng.random() * image.shape[1])
return r1, c1, r2, c2
while True:
r1, c1, r2, c2 = sample()
if (r1 - r2)**2 + (c1 - c2)**2 > 25:
break
rr, cc, val = line_aa(r1, c1, r2, c2)
line_mask = np.zeros_like(image)
line_mask[rr, cc] = val
line_mask = (gaussian(line_mask, sigma=3) > 0.01)
return line_mask
def draw_random_line_aa(im, red, green, blue):
shape = im.shape
row_length = shape[1]
column_length = shape[2]
r0 = np.random.randint(0, high=row_length)
c0 = np.random.randint(0, high=column_length)
r1 = np.random.randint(0, high=row_length)
c1 = np.random.randint(0, high=column_length)
rr, cc, val = line_aa(r0, c0, r1, c1)
im[0, rr, cc] = val * red
im[1, rr, cc] = val * green
im[2, rr, cc] = val * blue
return im
def draw_pols_as_lines(pols, sz = [640,640]): ''' Inputs: Accepts list of polygons and a size tuple/list for output image. Outputs: Boolean Image of polygons lines drawn Example: >> POLS = [ [[404, 236], [496, 357], [304, 492]], [[125, 270], [157, 307], [131, 323]], ] >> draw_pols_as_lines(POLS, sz=[640,640]) ''' img = np.zeros(sz) for pol in pols: if pol: for i in range(len(pol)): p0, p1 = pol[i], pol[(i+1)%len(pol)] rr, cc, val = draw.line_aa(int(p0[0]), int(p0[1]), int(p1[0]), int(p1[1])) img[rr, cc] = 1 return img.astype(bool)
def drawline(theta, row_c, col_c, row, col):
radians = math.radians(theta)
tang = math.tan(radians)
row_t = []
col_t = []
if theta == 45:
row_t.append(0)
row_t.append(row - 1)
col_t.append(col - 1)
col_t.append(0)
if theta == 135:
row_t.append(0)
row_t.append(row - 1)
col_t.append(0)
col_t.append(col - 1)
if theta is 90:
row_t.append(0)
row_t.append(row - 1)
col_t.append(col_c)
col_t.append(col_c)
elif theta is 0:
row_t.append(row_c)
row_t.append(row_c)
col_t.append(0)
col_t.append(col - 1)
else:
row_V = tang * (col - col_c)
col_V = (float(row_c)) / float(tang)
if abs(row_V) > row_c:
row_t.append(0)
row_t.append(row - 1)
col_t.append(col_c + int(col_V))
col_t.append(col_c - int(col_V))
elif abs(col_V) > col - col_c:
col_t.append(0)
col_t.append(col - 1)
row_t.append(row_c + int(row_V))
row_t.append(row_c - int(row_V))
rr, cc, val = line_aa(row_t[0], col_t[0], row_t[1], col_t[1])
return rr, cc
def draw_lines(lines, sz=[640, 640]):
"""
Inputs: Accepts list of lines, a line is a tuple, which contains two tuples,
point A and point B. A size sz which is default to (640,640)
Outputs: A Boolean Image of Lines,
Example:
>> lines = [
((10,10),(20,20)),
((30,30),(50,50))
]
>> draw_lines(lines, sz = (100,100))
"""
img = np.zeros(sz)
for line in lines:
p1, p2 = line
r0, c0 = p1
r1, c1 = p2
rr, cc, val = draw.line_aa(int(c0), int(r0), int(c1), int(r1))
img[rr, cc] = 1
return img.astype(bool)
def draw_pose_from_cords_and_visibility(pose_joints, visibility, img_size, radius=2, draw_joints=True):
colors = np.zeros(shape=img_size + (3,), dtype=np.uint8)
if draw_joints:
for f, t in LIMB_SEQ:
if _is_invalid(pose_joints[f], visibility[f]) or _is_invalid(pose_joints[t], visibility[t]):
continue
yy, xx, val = line_aa(pose_joints[f][0], pose_joints[f][1], pose_joints[t][0], pose_joints[t][1])
try:
colors[yy, xx] = np.expand_dims(val, 1) * 255
except IndexError:
print(pose_joints[f], pose_joints[t])
for i, joint in enumerate(pose_joints):
if _is_invalid(pose_joints[i], visibility[i]):
continue
yy, xx = circle(joint[0], joint[1], radius=radius, shape=img_size)
colors[yy, xx] = COLORS[i]
return colors
def draw_line(canvas, start, end, value=0, opacity=.6, fast=False):
"""
In place draws a line with anti-aliasing and given opacity.
canvas: array where to apply the line in-place.
start: Initial row and column.
end: Final row and column.
value: color to paint.
opacity: how much to affect the image with the new values.
fast: it'll would execute a single operation, this ignores anti-alianing.
"""
if fast:
rr, cc = draw.line(*start, *end)
original = canvas[rr, cc]
canvas[rr, cc] = value * opacity + original * (1 - opacity)
else:
rr, cc, alpha = draw.line_aa(*start, *end)
original = canvas[rr, cc]
painted = value * alpha + original * (1 - alpha)
canvas[rr, cc] = painted * opacity + original * (1 - opacity)
def draw_pose_from_cords(pose_joints, img_size, target_img, radius=2, draw_joints=True):
colors = np.zeros(shape=img_size + (3, ), dtype=np.uint8)
colors = target_img
if draw_joints:
for f, t in LIMB_SEQ:
from_missing = pose_joints[f][0] == MISSING_VALUE or pose_joints[f][1] == MISSING_VALUE
to_missing = pose_joints[t][0] == MISSING_VALUE or pose_joints[t][1] == MISSING_VALUE
if from_missing or to_missing:
continue
yy, xx, val = line_aa(pose_joints[f][0], pose_joints[f][1], pose_joints[t][0], pose_joints[t][1])
colors[yy, xx] = np.expand_dims(val, 1) * 255
for i, joint in enumerate(pose_joints):
if pose_joints[i][0] == MISSING_VALUE or pose_joints[i][1] == MISSING_VALUE:
continue
yy, xx = circle(joint[0], joint[1], radius=radius, shape=img_size)
colors[yy, xx] = COLORS[i]
return colors
def filter_contours_lines(v0, v1, lines, img, dist): #linie zaczynajace sie w danym punkcie w otoczeniu dist pikseli
print "v0, v1:" , v0, v1
max_line_length = 0
for tmp_line in lines:
tmp0, tmp1 = tmp_line
tp0 = (tmp0[1], tmp0[0])
tp1 = (tmp1[1], tmp1[0])
if distance(tmp0, tmp1) > max_line_length:
max_line_length = distance(tmp0, tmp1)
print "NEW_MAX:", tmp0, tmp1
print "tp0 tp1: " ,tp0, tp1
print "v0 tp0", distance(v0, tp0)
print "v0 tp1", distance(v0, tp1)
print "v1 tp0", distance(v1, tp0)
print "v1 tp1", distance(v1, tp1)
if distance(v0, tp0) < dist or distance(v0, tp1) < dist or distance(v1, tp0) < dist or distance(v1, tp1) < dist:
print "removing line"
rr, cc, v = line_aa(tmp0[1], tmp0[0], tmp1[1], tmp1[0])
img[rr,cc] = 0
return img, max_line_length
def draw_pose_from_cords(pose_joints,
observed,
img_size=(256, 256),
radius=2,
draw_joints=True,
img=np.zeros([224, 224, 3])):
pose_joints = np.round(pose_joints).astype(np.int32)
pose_joints[pose_joints < 0] = 0
pose_joints[pose_joints > 255] = 255
pose_joints = pose_joints.reshape([18, 2])
colors = np.zeros(shape=img_size + (3, ), dtype=np.uint8)
# interesting case of numpy array assignment semantics here . . the below ass is a view, hence the overliad poses error that destroyed the precious time of an incomporable specimen on this overpopulated information craving filth obsessed world
# colors = img
mask = np.zeros(shape=img_size, dtype=bool)
if draw_joints:
for f, t in LIMB_SEQ:
from_missing = pose_joints[f][0] == MISSING_VALUE or pose_joints[
f][1] == MISSING_VALUE or observed[f] == 0 or observed[18 +
f] == 0
to_missing = pose_joints[t][0] == MISSING_VALUE or pose_joints[t][
1] == MISSING_VALUE or observed[t] == 0 or observed[18 +
t] == 0
if from_missing or to_missing:
continue
yy, xx, val = line_aa(pose_joints[f][0], pose_joints[f][1],
pose_joints[t][0], pose_joints[t][1])
colors[yy, xx] = np.expand_dims(val, 1) * 255
mask[yy, xx] = True
for i, joint in enumerate(pose_joints):
if pose_joints[i][0] == MISSING_VALUE or pose_joints[i][
1] == MISSING_VALUE or observed[i] == 0 or observed[18 +
i] == 0:
continue
yy, xx = circle(joint[0], joint[1], radius=radius, shape=img_size)
colors[yy, xx] = COLORS[i]
mask[yy, xx] = True
# colors = colors.transpose((2,0,1))
return colors
def draw_shape(img, shape):
'''
Takes in shapes as input and
:param img: image to modify
:param shape: shape to draw ( a series of x and y paired coordinates)
:return: a new image that has the shape drawn
'''
from skimage.draw import line_aa
# print("Drawing the next shape")
for i in range(len(shape) - 1):
# print("First shape x and y : " + str(shape[i]))
# print("Second shape x and y : " + str(shape[i+1]))
rr, cc, val = line_aa(shape[i][0], shape[i][1], shape[i + 1][0],
shape[i + 1][1])
# print(val)
# print(rr)
# print(cc)
img[rr, cc] = val * 255
return img
def getDisplacementMap(self, xx, yy, steps, scale, mask): xx = sitk.GetArrayFromImage(xx) yy = sitk.GetArrayFromImage(yy) dx = int(self.frameSize[0] / steps) dy = int(self.frameSize[1] / steps) dmap = np.ones( (self.frameSize[0], self.frameSize[1]), dtype = 'uint8' ) * 255 for x in range(0, self.frameSize[0], dx): for y in range(0, self.frameSize[1], dy): if mask[y,x] > 0: rr,cc,val = circle_perimeter_aa( y, x, 2 ) print(rr,cc) dmap[rr,cc] = 255 - (val * 58) rr,cc,val = line_aa( y, x, int(y + yy[y,x] * scale), int(x + xx[y,x] * scale) ) print(rr,cc) dmap[rr,cc] = 255- (val * 255) return dmap
def gen_scored_points():
for points in itertools.combinations(intersection_points, 4):
points = list(map(np.array, sorted(points, key=lambda x: -sum(x))))
sorted_points = [points.pop()[:2]]
while len(points) > 0:
points.sort(key=lambda x: -min(
np.abs(np.array(x) - sorted_points[-1])))
sorted_points.append(points.pop())
sorted_points = [np.round(p).astype(int) for p in sorted_points]
score = 0
deltas = [sorted_points[0] - sorted_points[-1]]
for i in range(4):
p1 = sorted_points[i]
p2 = sorted_points[(i + 1) % 4]
delta = p2 - p1
# filter out small edges
# print('|delta|=',np.linalg.norm(delta))
if np.linalg.norm(delta) < min(im_rows, im_cols) / 2:
score = 0
break
# filter out acute angles
angle = np.degrees(
np.arccos(delta @ deltas[-1] / np.linalg.norm(delta) /
np.linalg.norm(deltas[-1])))
# print('angle=', angle)
if angle < 60 or angle > 120:
score = 0
break
rr, cc, vals = line_aa(p1[1], p1[0], p2[1], p2[0])
ix_1 = rr < im_rows - 1
ix_2 = cc < im_cols - 1
ix = ix_1 * ix_2
score += np.sum(vals[ix] * original_edged[rr[ix], cc[ix]])
deltas.append(delta)
yield score, sorted_points, deltas
def visualize(sketch):
assert sketch.shape[1] in [3, 5]
# start token
if sketch.shape[1] == 3:
sketch = np.concatenate(([[0, 0, 0]], sketch))
elif sketch.shape[1] == 5:
sketch = np.concatenate(([[0, 0, 1, 0, 0]], sketch))
length = np.where(sketch[:, 4] == 1)[0]
if len(length) > 0:
sketch = sketch[:length[0] + 1]
# coordinates
sketch[:, 0] = np.cumsum(sketch[:, 0], 0)
sketch[:, 0] -= np.min(sketch[:, 0])
sketch[:, 1] = np.cumsum(sketch[:, 1], 0)
sketch[:, 1] -= np.min(sketch[:, 1])
if np.max(sketch[:, :2]) > 0:
sketch[:, :2] /= np.max(sketch[:, :2])
sketch[:, :2] *= 255
# sequence => strokes
if sketch.shape[1] == 3:
strokes = np.split(sketch, np.where(sketch[:, 2] == 1)[0] + 1)
elif sketch.shape[1] == 5:
strokes = np.split(sketch, np.where(sketch[:, 3] == 1)[0] + 1)
# canvas
canvas = np.ones((256, 256, 3), dtype = np.float32)
index, length = 0, np.sum([len(stroke) - 1 for stroke in strokes])
for stroke in strokes:
for k in range(len(stroke) - 1):
x1, y1 = map(int, stroke[k, :2])
x2, y2 = map(int, stroke[k + 1, :2])
rr, cc, _ = line_aa(y1, x1, y2, x2)
canvas[rr, cc, :] = .75 * index / length
index += 1
return canvas.transpose(2, 0, 1)
def custom_train_loop(self, sess, train_targets, **loop_params):
"""Define Custom training loop.
Args:
sess (tf.Session): Current tensorflow session.
train_targets (list): Description.
**loop_params: Optional kwargs needed to perform custom train loop.
Returns:
dict: A dictionary containing train targets evaluated by the session.
"""
# boxes = var_dict['boxes']
# boxes_val = sess.run(boxes)
# import pdb; pdb.set_trace()
max_obj = 0
for i in range(20):
# ih, iw, image, obj_count, boxes = sess.run([var_dict[k] for k in ['ih', 'iw', 'images', 'num_objects', 'boxes']]) #['images', 'labels',
ih, iw, obj_count, boxes, images = sess.run([
var_dict[k]
for k in ['ih', 'iw', 'num_objects', 'boxes', 'images']
])
max_obj = max(max_obj, obj_count)
print i, ih, iw, obj_count, boxes[
0][:obj_count[0]] #, image.shape, obj_count
img = np.array(images[0])
x_center, y_center, w, h = boxes[0, 0, :4]
coords = [(x_center - w / 2), (x_center + w / 2),
(y_center - h / 2), (y_center + h / 2)] # x1, x2, y1, y2
x1, x2, y1, y2 = [int(c) for c in coords]
print([int(c) for c in coords])
rr, cc, val = line_aa(y1, x1, y2, x2)
img[rr, cc, 0] = val
imsave('image_{}.png'.format(i), img)
import pdb
pdb.set_trace()
train_results, p = sess.run([train_targets, var_dict['print']])
for i, result in enumerate(train_results):
print('Model {} has loss {}'.format(i, result['loss']))
return train_results
def collect_line_coords(seg_line_df, scale=1):
# group according to line idx
grouped = seg_line_df.groupby('line_idx')
# collect all line coordinates
rr_list, cc_list, lbbox_list = [], [], []
for i, line_rec in grouped:
xx = np.rint(line_rec.x.values * scale).astype(int)
yy = np.rint(line_rec.y.values * scale).astype(int)
lbbox = np.array([np.min(xx), np.min(yy), np.max(xx), np.max(yy)])
lbbox_list.append(lbbox)
for li in range(len(xx) - 1):
rr, cc, _ = line_aa(yy[li], xx[li], yy[li + 1], xx[li + 1])
# rr, cc = line(yy[li], xx[li], yy[li+1], xx[li+1])
rr_list.append(rr)
cc_list.append(cc)
# stack coordinates
rr = np.hstack(rr_list)
cc = np.hstack(cc_list)
lbboxes = np.stack(lbbox_list)
return rr, cc, lbboxes
def draw_pose_from_cords(pose_joints, img_size, radius=2, draw_joints=True):
colors = np.zeros(shape=img_size + (3, ), dtype=np.uint8)
mask = np.zeros(shape=img_size, dtype=bool)
if draw_joints:
for f, t in LIMB_SEQ:
from_missing = MISSING(pose_joints[f][0]) or MISSING(
pose_joints[f][1])
to_missing = MISSING(pose_joints[t][0]) or MISSING(
pose_joints[t][1])
if from_missing or to_missing:
continue
'''
Trick, use a 4-polygon with 1 pixel width to represent lines, involve shape control.
yy, xx = polygon(
[pose_joints[f][0], pose_joints[t][0], pose_joints[t][0]+1, pose_joints[f][0]+1],
[pose_joints[f][1], pose_joints[t][1], pose_joints[t][1]+1, pose_joints[f][1]+1],
shape=img_size
)
'''
yy, xx, val = line_aa(pose_joints[f][0], pose_joints[f][1],
pose_joints[t][0], pose_joints[t][1])
valid_ids = [
i for i in range(len(yy))
if 0 < yy[i] < img_size[0] and 0 < xx[i] < img_size[1]
]
yy, xx, val = yy[valid_ids], xx[valid_ids], val[valid_ids]
colors[yy, xx] = np.expand_dims(val, 1) * 255
mask[yy, xx] = True
for i, joint in enumerate(pose_joints):
if MISSING(pose_joints[i][0]) or MISSING(pose_joints[i][1]):
continue
yy, xx = circle(joint[0], joint[1], radius=radius, shape=img_size)
colors[yy, xx] = COLORS[i]
mask[yy, xx] = True
return colors, mask
def simplify_polygon(pol, tol=2, max_iter=20):
"""
Inputs: Accepts a polygon, tolerance=2 and maximum iteration.
Default values are tol=2 and max_iter=20
Outputs: Simplified Polygon. Works kind of like measure.approximate_polygon.
Example:
>> POL = [
[125, 270], [157, 307], [131, 323]
]
>> simplify_polygon(POL)
"""
iiter = 0
while True:
iiter = iiter + 1
pol1 = []
can_skip = set()
i = 1
while i < len(pol) - 1:
p0, p1 = pol[i - 1], pol[i + 1]
rr, cc, val = draw.line_aa(int(p0[0]), int(p0[1]), int(p1[0]),
int(p1[1]))
pts = set(zip(rr, cc))
pi = set()
for x in range(int(int(pol[i][0]) - tol / 2),
int(int(pol[i][0]) + tol / 2) + 1):
for y in range(int(int(pol[i][1]) - tol / 2),
int(int(pol[i][1]) + tol / 2) + 1):
pi.add((x, y))
if len(pi.intersection(pts)) > 0:
can_skip.add(i)
i = i + 1
i = i + 1
if len(can_skip) == 0 or iiter >= max_iter:
break
for i in range(len(pol)):
if i not in can_skip:
pol1.append(pol[i])
pol = pol1
return pol
def draw_graph(self, data_vector, color):
#interpolate the data vector to fill in gaps
d_interpld = interp1d(self.state_x, data_vector)
#convert data vector to a data array the size of the window's x dimension
data_bar = np.array(
[-d_interpld(x) * self.bhh + 0.5 for x in self.window_x],
dtype=np.int)
# All locations where we need to draw lines
data_jump_locs = []
for loc in np.where(abs(data_bar[:-1] - data_bar[1:]) >= 2)[0]:
rr, cc, val = line_aa(data_bar[loc] + self.fh + self.bhh, loc,
data_bar[loc + 1] + self.fh + self.bhh,
loc + 1)
data_jump_locs.append((rr, cc))
""" Draw the speed and steering lines on the bar below the video.
Takes about 1ms to run through the for loops"""
self.window[data_bar + self.fh + self.bhh, self.fwi, :] = color
for rr, cc in data_jump_locs:
self.window[rr, cc, :] = color
def spatial_draw(self, image_size, pose_joints, joint_line_list, radius=2, line_color=[118,214,255], cycle_color=[66,115,177]):
colors = np.ones(shape=image_size + (3, ), dtype=np.uint8)*255.0
mask = np.zeros(shape=image_size, dtype=np.uint8)
for f, t in joint_line_list:
yy, xx, val = line_aa(pose_joints[0,f], pose_joints[1,f], pose_joints[0,t], pose_joints[1,t])
yy[yy>image_size[0]-1]=image_size[0]-1
xx[xx>image_size[1]-1]=image_size[1]-1
mask[yy, xx] = 1
mask = morphology.dilation(mask, morphology.disk(radius=1))
colors[mask==1] = line_color
mask = np.zeros(shape=image_size, dtype=np.uint8)
for i in range(pose_joints.shape[1]):
yy, xx, val = circle_perimeter_aa(pose_joints[0,i], pose_joints[1,i], radius=radius)
# yy, xx = circle(pose_joints[0,i], pose_joints[1,i], radius=radius, shape=im_size)
yy[yy>image_size[0]-1]=image_size[0]-1
xx[xx>image_size[1]-1]=image_size[1]-1
mask[yy, xx] = 1
# mask = morphology.dilation(mask, morphology.disk(radius=1))
colors[mask==1] = cycle_color
return colors
def drawLine(self, centerX, centerY, angle, length, colour):
"""Plot a line from an angle, length and center
Parameters:
colour: 0:1 intensity percentage
Attributes:
"""
# Calculate the offset of the start and end from the center
radAngle = math.radians(angle)
xOffset = length * math.cos(radAngle)
yOffset = length * math.sin(radAngle)
# Calculate start and finish of the lines
startx = math.floor(centerX - xOffset)
stopx = math.floor(centerX + xOffset)
starty = math.floor(centerY - yOffset)
stopy = math.floor(centerY + yOffset)
# Draw the line onto the array
rr, cc, val = line_aa(starty, startx, stopy, stopx)
#rr, cc = line(starty, startx, stopy, stopx)
self._outputImage[rr, cc] = colour * 255
def generate_canvas(rhos, thetas, frame):
"""
From every line segment a line is extended in either direction. This is line is then drawn on canvas.
Based on the line density(i.e. number of intersections) we can identify a vanishing point
:param rhos:
:param thetas:
:param frame:
:return: canvas
"""
h, w, c = frame.shape
canvas = np.zeros((h * 5, w * 5), np.int)
for rho, theta in zip(rhos, thetas):
a = np.cos(theta)
b = np.sin(theta)
rho = rho + (2. * h) * b + (2. * w) * a
if b == 0:
x1, x2 = rho / a, (rho - h * 5 * b) / a
y1, y2 = rho, rho
elif a == 0:
y1, y2 = rho / b, (rho - w * 5 * a) / b
x1, x2 = rho, rho
else:
x1, x2, y1, y2 = rho / a, (rho - h * 5 * b) / a, rho / b, (
rho - w * 5 * a) / b
coordinates = []
if 0 <= x1 < w * 5:
coordinates.append((int(x1), 0))
if 0 <= x2 < w * 5:
coordinates.append((int(x2), int(h * 5 - 1)))
if 0 <= y1 < h * 5:
coordinates.append((0, int(y1)))
if 0 <= y2 < h * 5:
coordinates.append((int(w * 5 - 1), (int(y2))))
[(x1, y1), (x2, y2)] = coordinates
rr, cc, val = line_aa(y1, x1, y2, x2)
canvas[rr, cc] = canvas[rr, cc] + 1
return canvas
def graph_image(self, xs, ys):
"""
creates an image that represents the graph structure of connected nodes
used for plotting as an image, so that the scale is the same as the occupancy images
"""
img = np.zeros((len(xs), len(ys)))
for n in self.nodes:
x = n.data['location'][0]
y = n.data['location'][1]
# Index of the coordinate in the array
# NOTE: this might be slightly incorrect....?
ix = (np.abs(xs - x)).argmin()
iy = (np.abs(ys - y)).argmin()
img[ix, iy] = 1 #TEMP
# Go through the neighbor indices of each neighbor
for ni in n.get_neighbors():
# Draw a line between the current node and the neighbor node
nx = self.nodes[ni].data['location'][0]
ny = self.nodes[ni].data['location'][1]
inx = (np.abs(xs - nx)).argmin()
iny = (np.abs(ys - ny)).argmin()
# rr, cc = line(ix, iy, inx, iny)
# anti-aliased line
rr, cc, val = line_aa(ix, iy, inx, iny)
img[rr, cc] = val
# print(np.max(img))
return img
def line(img, coord, color):
"""
Draws a line on the input image
Parameters
----------
img: numpy.ndarray
input image with shape [nb_lin, nb_col, nb_channels]
coord: tuple of integers with format (y1, x1, y2, x2)
position (before rotation) and size of the rectangle
color: array like type
Rectangle color. Size of array must be equal to nb_channels of
input img
Returns
-------
out numpy.ndarray
img with drawn rectangle
"""
# Line coordinates
y1, x1, y2, x2 = coord
# Generate line coordinates
rr, cc, val = line_aa(y1, x1, y2, x2)
# Draw line color
out = img.copy()
val = val[:, None]
color = np.asarray(color)[None, :]
print "(out[rr, cc] * (1 - val)).shape = ", (out[rr, cc] * (1 - val)).shape
print "(out[rr, cc] * val * color).shape = ", (out[rr, cc] * val *
color).shape
out[rr, cc] = out[rr, cc] * (1 - val) + out[rr, cc] * val * color
return out
def line(self,
row1: int,
col1: int,
row2: int,
col2: int,
color: ColorType = None,
alpha: float = 1.0) -> 'Layer':
"""
Draw a line between two points
:param row1: Start row
:param col1: Start column
:param row2: End row
:param col2: End column
:param color: Color to draw with
"""
rr, cc, aa = draw.line_aa(clamp(0, self.height, row1),
clamp(0, self.width, col1),
clamp(0, self.height, row2),
clamp(0, self.width, col2))
self._draw(rr, cc, color, aa)
return self
def hog(image, orientations=9, pixels_per_cell=(8, 8),
cells_per_block=(2, 2), visualise=True, normalise=False):
"""Extract Histogram of Oriented Gradients (HOG) for a given image.
Compute a Histogram of Oriented Gradients (HOG) by
1. (optional) global image normalisation
2. computing the gradient image in x and y
3. computing gradient histograms
4. normalising across blocks
5. flattening into a feature vector
Parameters
----------
image : (M, N) ndarray
Input image (greyscale).
orientations : int
Number of orientation bins.
pixels_per_cell : 2 tuple (int, int)
Size (in pixels) of a cell.
cells_per_block : 2 tuple (int,int)
Number of cells in each block.
visualise : bool, optional
Also return an image of the HOG.
normalise : bool, optional
Apply power law compression to normalise the image before
processing.
Returns
-------
newarr : ndarray
HOG for the image as a 1D (flattened) array.
hog_image : ndarray (if visualise=True)
A visualisation of the HOG image.
References
----------
* http://en.wikipedia.org/wiki/Histogram_of_oriented_gradients
* Dalal, N and Triggs, B, Histograms of Oriented Gradients for
Human Detection, IEEE Computer Society Conference on Computer
Vision and Pattern Recognition 2005 San Diego, CA, USA
"""
image = np.atleast_2d(image)
"""
The first stage applies an optional global image normalisation
equalisation that is designed to reduce the influence of illumination
effects. In practice we use gamma (power law) compression, either
computing the square root or the log of each colour channel.
Image texture strength is typically proportional to the local surface
illumination so this compression helps to reduce the effects of local
shadowing and illumination variations.
"""
if image.ndim > 3:
raise ValueError("Currently only supports grey-level images")
if normalise:
image = sqrt(image)
"""
The second stage computes first order image gradients. These capture
contour, silhouette and some texture information, while providing
further resistance to illumination variations. The locally dominant
colour channel is used, which provides colour invariance to a large
extent. Variant methods may also include second order image derivatives,
which act as primitive bar detectors - a useful feature for capturing,
e.g. bar like structures in bicycles and limbs in humans.
"""
gx = np.zeros(image.shape)
gy = np.zeros(image.shape)
gx[:, :-1] = np.diff(image, n=1, axis=1)
gy[:-1, :] = np.diff(image, n=1, axis=0)
"""
The third stage aims to produce an encoding that is sensitive to
local image content while remaining resistant to small changes in
pose or appearance. The adopted method pools gradient orientation
information locally in the same way as the SIFT [Lowe 2004]
feature. The image window is divided into small spatial regions,
called "cells". For each cell we accumulate a local 1-D histogram
of gradient or edge orientations over all the pixels in the
cell. This combined cell-level 1-D histogram forms the basic
"orientation histogram" representation. Each orientation histogram
divides the gradient angle range into a fixed number of
predetermined bins. The gradient magnitudes of the pixels in the
cell are used to vote into the orientation histogram.
"""
magnitude = sqrt(gx ** 2 + gy ** 2)
orientation = arctan2(gy, (gx + 1e-15)) * (180 / pi) + 90
sy, sx = image.shape
cx, cy = pixels_per_cell
bx, by = cells_per_block
n_cellsx = int(np.floor(sx // cx)) # number of cells in x
n_cellsy = int(np.floor(sy // cy)) # number of cells in y
# compute orientations integral images
orientation_histogram = np.zeros((n_cellsy, n_cellsx, orientations))
for i in range(orientations):
#create new integral image for this orientation
# isolate orientations in this range
temp_ori = np.where(orientation < 180 / orientations * (i + 1),
orientation, 0)
temp_ori = np.where(orientation >= 180 / orientations * i,
temp_ori, 0)
# select magnitudes for those orientations
cond2 = temp_ori > 0
temp_mag = np.where(cond2, magnitude, 0)
orientation_histogram[:,:,i] = uniform_filter(temp_mag, size=(cy, cx))[cy/2::cy, cx/2::cx]
# now for each cell, compute the histogram
#orientation_histogram = np.zeros((n_cellsx, n_cellsy, orientations))
radius = min(cx, cy) // 2 - 1
hog_image = None
if visualise:
hog_image = np.zeros((sy, sx), dtype=float)
if visualise:
from skimage import draw
for x in range(n_cellsx):
for y in range(n_cellsy):
for o in range(orientations):
centre = tuple([y * cy + cy // 2, x * cx + cx // 2])
dx = radius * cos(float(o) / orientations * np.pi)
dy = radius * sin(float(o) / orientations * np.pi)
rr, cc, val = draw.line_aa(centre[0] - int(dx), centre[1] - int(dy), centre[0] + int(dx), centre[1] + int(dy))
hog_image[rr, cc] += orientation_histogram[y, x, o]
"""
The fourth stage computes normalisation, which takes local groups of
cells and contrast normalises their overall responses before passing
to next stage. Normalisation introduces better invariance to illumination,
shadowing, and edge contrast. It is performed by accumulating a measure
of local histogram "energy" over local groups of cells that we call
"blocks". The result is used to normalise each cell in the block.
Typically each individual cell is shared between several blocks, but
its normalisations are block dependent and thus different. The cell
thus appears several times in the final output vector with different
normalisations. This may seem redundant but it improves the performance.
We refer to the normalised block descriptors as Histogram of Oriented
Gradient (HOG) descriptors.
"""
n_blocksx = (n_cellsx - bx) + 1
n_blocksy = (n_cellsy - by) + 1
normalised_blocks = np.zeros((n_blocksy, n_blocksx,
by, bx, orientations))
for x in range(n_blocksx):
for y in range(n_blocksy):
block = orientation_histogram[y:y + by, x:x + bx, :]
eps = 1e-5
normalised_blocks[y, x, :] = block / sqrt(block.sum() ** 2 + eps)
"""
The final step collects the HOG descriptors from all blocks of a dense
overlapping grid of blocks covering the detection window into a combined
feature vector for use in the window classifier.
"""
if visualise:
return normalised_blocks.ravel(), hog_image
else:
return normalised_blocks.ravel()
img[rr, cc, :] = (1, 0, 1)
rr, cc = ellipse_perimeter(120, 400, 60, 20, orientation=-math.pi / 4.)
img[rr, cc, :] = (0, 0, 1)
rr, cc = ellipse_perimeter(120, 400, 60, 20, orientation=math.pi / 2.)
img[rr, cc, :] = (1, 1, 1)
ax1.imshow(img)
ax1.set_title('No anti-aliasing')
ax1.axis('off')
from skimage.draw import line_aa, circle_perimeter_aa
img = np.zeros((100, 100), dtype=np.double)
# anti-aliased line
rr, cc, val = line_aa(12, 12, 20, 50)
img[rr, cc] = val
# anti-aliased circle
rr, cc, val = circle_perimeter_aa(60, 40, 30)
img[rr, cc] = val
ax2.imshow(img, cmap=plt.cm.gray, interpolation='nearest')
ax2.set_title('Anti-aliasing')
ax2.axis('off')
plt.show()
def fast_triangle_mesh_drawer(tris, origin, look):
# shouldn't do anything
look /= norm(look)
img = np.zeros((Y, X))
zimg = np.ones((Y, X))*np.inf
def project_point(pt):
# vector from pt to origin
vv = pt - origin
v = vv/norm(vv)
# real projection shit
vx = npa((v[0], 0, v[2]))
lx = npa((look[0], 0, look[2]))
vy = npa((0, v[1], v[2]))
ly = npa((0, look[1], look[2]))
def ang(v1, v2):
v1 /= norm(v1)
v2 /= norm(v2)
angl = np.dot(v1, v2)
crs = np.cross(v1, v2)
if np.sum(crs) >= 0.0:
return np.arccos(angl)
else:
return -np.arccos(angl)
x = (ang(vx, lx) / arcrad_per_pixel)
y = (ang(vy, ly) / arcrad_per_pixel)
# add z for z-buffering
# z is the distance of the point from the plane formed by look and origin
# project v on to look
z = np.dot(v, look) * norm(vv)
"""
print " *** "
print v, K
print pt, x, y
"""
return int(round(x + X/2)),int(round(Y/2 - y)), z
# project the triangles into 2D space
# does this projection preserve the u and v
DRAW_WIREFRAME = False
if DRAW_WIREFRAME:
lines = []
for tr in tris:
p0, p1, p2 = project_point(tr[0]), project_point(tr[1]), project_point(tr[2])
lines.append((p0[0:2], p1[0:2]))
lines.append((p1[0:2], p2[0:2]))
lines.append((p2[0:2], p0[0:2]))
for pt1, pt2 in lines:
rr, cc, val = line_aa(pt1[1], pt1[0], pt2[1], pt2[0])
# filter
rr[np.logical_or(rr < 0, rr >= Y)] = 0
cc[np.logical_or(cc < 0, cc >= X)] = 0
img[rr, cc] = val
else:
# z-buffering, keeping the quality low in line with the rest of the program
polys = []
for tr in tris:
xyz = npa(map(project_point, tr))
# get min and max
xmin, xmax = np.min(xyz[:, 0]), np.max(xyz[:, 0])
ymin, ymax = np.min(xyz[:, 1]), np.max(xyz[:, 1])
# on screen
xmin = np.clip(xmin, 0, X).astype(np.int)
xmax = np.clip(xmax, 0, X).astype(np.int)
ymin = np.clip(ymin, 0, Y).astype(np.int)
ymax = np.clip(ymax, 0, Y).astype(np.int)
# triangle in 3 space
vs1 = xyz[1][0:2] - xyz[0][0:2]
vs2 = xyz[2][0:2] - xyz[0][0:2]
vsx = np.cross(vs1, vs2)
# shade
shade = random.uniform(0.3, 1.0)
for x in range(xmin, xmax):
for y in range(ymin, ymax):
q = npa([x,y]) - xyz[0][0:2]
u = np.cross(q, vs2) / vsx
v = np.cross(vs1, q) / vsx
if u >= 0 and v >= 0 and u+v <= 1.0:
pt_xyz = (1-u-v)*xyz[0] + u*xyz[1] + v*xyz[2]
if pt_xyz[2] < zimg[y,x]:
zimg[y,x] = pt_xyz[2]
img[y,x] = shade
#polys.append((np.mean(xyz[:, 2]), xyz))
"""
polys = sorted(polys, reverse=True, key=lambda x: x[0])
for _, xyz in polys:
rr, cc = polygon(xyz[:, 1], xyz[:, 0])
rr[np.logical_or(rr < 0, rr >= Y)] = 0
cc[np.logical_or(cc < 0, cc >= X)] = 0
img[rr, cc] = random.uniform(0.3, 1.0)
"""
return img
#!/usr/bin/env python
import numpy as np
import matplotlib.pyplot as plt
from skimage.transform import hough_line, hough_line_peaks
from skimage.draw import line_aa
if __name__ == '__main__':
# Construct test image
image = np.zeros((200, 100))
# Line 0
rr, cc, val = line_aa(50, 15, 140, 97)
image[rr, cc] = val
# Line 1
rr, cc, val = line_aa(100, 5, 12, 84)
image[rr, cc] = np.fmax(image[rr, cc], val)
# Line 2
rr, cc, val = line_aa(140, 15, 160, 80)
image[rr, cc] = np.fmax(image[rr, cc], val)
# Compute the Hough transform for display
h, theta, d = hough_line(image)
fig, ax = plt.subplots(1, 3, figsize=(6, 3.5))
# Original image
ax[0].imshow(image, cmap=plt.cm.gray)
def add_line(arr, p1x, p1y, p2x, p2y):
from skimage.draw import line_aa
rr, cc, val = line_aa(p1y, p1x, p2y, p2x)
arr[rr, cc] = val * 255
return arr
