def _process(self, image, cnt_3d):
if len(cnt_3d) > 0:
cnt = to_2D_contour(cnt_3d)
# Determine the bounding rectangle of all contours
x, y, w, h = cv2.boundingRect(np.concatenate(cnt))
image_height, image_width = image.shape[0], image.shape[1]
# subtract the offset to move the bottom-left of the contour to [0,0]
cnt_3d = [
np.subtract(c, [x, y, 0], dtype=np.int32) for c in cnt_3d
]
preview_image = np.zeros(image.shape, dtype="uint8")
preview_image.fill(255)
preview_cnt = contour_into_image(to_2D_contour(cnt_3d),
preview_image)
x, y, w, h = cv2.boundingRect(np.concatenate(preview_cnt))
# draw the coordinate axes
# x-axis
cv2.line(preview_image, (x, y + h), (x + w, y + h), (255, 0, 0), 1)
# y-axis
cv2.line(preview_image, (x, y), (x, y + h), (0, 0, 255), 1)
# draw the contour itself
cv2.drawContours(preview_image, preview_cnt, -1, (60, 169, 242), 1)
image = preview_image
return image, cnt_3d
def _process(self, image, cnt_3d):
if len(cnt_3d) > 0:
cnt = to_2D_contour(cnt_3d)
# Determine the bounding rectangle of all contours
x, y, w, h = cv2.boundingRect(np.concatenate(cnt))
image_height, image_width = image.shape[0], image.shape[1]
# the offset to move the center of the contour to [0,0]
offset_x = int(w / 2 + x)
offset_y = int(h / 2 + y)
cnt_3d = [np.subtract(c, [offset_x, offset_y, 0], dtype=np.int32) for c in cnt_3d]
preview_image = np.zeros(image.shape, dtype="uint8")
preview_image.fill(255)
# generate a preview contour
#
preview_cnt = contour_into_image(to_2D_contour(cnt_3d), preview_image)
# draw the coordinate system of the centered drawing contour
x, y, w, h = cv2.boundingRect(np.concatenate(preview_cnt))
cv2.drawContours(preview_image, preview_cnt, -1, (60, 169, 242), 1)
# horizontal
cv2.line(preview_image, (x + int(w / 2), y + int(h / 2)), (x + w, y + int(h / 2)), (255, 0, 0), 1)
# vertical
cv2.line(preview_image, (x + int(w / 2), y), (x + int(w / 2), y + int(h / 2)), (0, 0, 255), 1)
image = preview_image
return image, cnt_3d
def processed(self, image, cnt):
self.cnt3D = cnt
if len(cnt) == 0:
return
# flip them upside down (gcode coordinate system vs. openCV coordinate system)
cnt2D = to_2D_contour(cnt)
x, y, w, h = cv2.boundingRect(np.concatenate(cnt2D))
transform_y = lambda y_coord: (-(y_coord - y) + (y + h))
cnt = ([
np.array([[p[0], transform_y(p[1]), p[2]] for p in c]) for c in cnt
])
itm_cnt = len(self.glWidget.items)
idx = itm_cnt - 1
while idx >= 0:
if type(self.glWidget.items[idx]) == gl.GLLinePlotItem:
del self.glWidget.items[idx]
idx -= 1
feed_rate = self.filter.feed_rate
clearance = self.filter.clearance
# Travel Toolpaths
first_up = [0, 0, clearance]
first_down = [0, 0, 0]
plt = gl.GLLinePlotItem(pos=np.array([[0, 0, 0], [50, 0, 0]]),
width=4,
antialias=True,
color="0000ff")
self.glWidget.addItem(plt)
plt = gl.GLLinePlotItem(pos=np.array([[0, 0, 0], [0, 50, 0]]),
width=4,
antialias=True,
color="FF0000")
self.glWidget.addItem(plt)
# Carving Toolpaths
for c in cnt:
c = c / 1000
plt = gl.GLLinePlotItem(pos=c,
width=2,
antialias=True,
color="ECB151")
self.glWidget.addItem(plt)
# Tool Movement to start
next_down = c[:1][0].tolist() # first element
next_up = [next_down[0], next_down[1], clearance] # first element
path = np.array([first_down, first_up, next_up, next_down])
plt = gl.GLLinePlotItem(pos=path,
width=1,
antialias=True,
color="3f3f3f3A")
self.glWidget.addItem(plt)
first_down = c[-1:][0].tolist()
first_up = [first_down[0], first_down[1], clearance]
def _process(self, image, cnt):
try:
single_channel = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
generated_cnt = []
row_index = 0 # [micro m]
normalized_depth = self.depth_in_micro_m / 255
for row in single_channel:
rle_row = [rle for rle in self.rle(row) if rle[0] < 250]
last_end = None
last_contour = None
for rle in rle_row:
# gray => depth in [micro m]
# 255 => 0 [micro m] # white means no cutting
# 0 => self.depth_in_micro_m [micro m] # black is full depth
depth = -(self.depth_in_micro_m - (normalized_depth * rle[0]))
gray_start = rle[1]
gray_end = rle[2] + gray_start
if gray_start == last_end:
last_contour = last_contour+[
[gray_start, row_index, depth],
[gray_end, row_index, depth]
]
else:
if not last_contour is None:
generated_cnt.append(np.array(last_contour, dtype=np.int32))
last_contour = [
[gray_start, row_index, depth],
[gray_end, row_index, depth]
]
last_end = gray_end
if not last_contour is None:
generated_cnt.append(np.array(last_contour, dtype=np.int32))
row_index += 1
# generate a preview image
#
preview_image = np.zeros(image.shape, dtype="uint8")
preview_image.fill(255)
# generate a preview contour
#
preview_cnt = contour_into_image(to_2D_contour(generated_cnt), preview_image)
cv2.drawContours(preview_image, preview_cnt, -1, (60, 169, 242), 1)
# draw the carving depth
cv2.putText(preview_image, "Carving Depth {:.2f} mm".format(self.depth_in_micro_m / 1000), (20, 50),
cv2.FONT_HERSHEY_SIMPLEX, 0.8, (255, 0, 0), 4)
image = preview_image
return image, generated_cnt
except Exception as exc:
exc_type, exc_obj, exc_tb = sys.exc_info()
fname = os.path.split(exc_tb.tb_frame.f_code.co_filename)[1]
print(exc_type, fname, exc_tb.tb_lineno)
print(type(self), exc)
def _process(self, image, cnt_3d):
try:
if len(cnt_3d) > 0:
cnt = to_2D_contour(cnt_3d)
x, y, w, h = cv2.boundingRect(np.concatenate(cnt))
image_height, image_width = image.shape[0], image.shape[1]
# the offset to move the center of the contour to [0,0]
cx = int(w / 2 + x)
cy = int(h / 2 + y)
a = math.radians(self.angle_in_degree)
ca = math.cos(a)
sa = math.sin(a)
rotate_point = lambda p: [
int(((p[0] - cx) * ca) - ((p[1] - cy) * sa) + cx),
int(((p[0] - cx) * sa) + ((p[1] - cy) * ca) + cy),
p[2]
]
rotate_cnt = lambda c: np.array([rotate_point(p) for p in c])
cnt_3d = [rotate_cnt(c) for c in cnt_3d]
preview_image = np.zeros(image.shape, dtype="uint8")
preview_image.fill(255)
# generate a preview contour
#
preview_cnt = contour_into_image(to_2D_contour(cnt_3d), preview_image)
# draw the coordinate system of the centered drawing contour
x, y, w, h = cv2.boundingRect(np.concatenate(preview_cnt))
cv2.drawContours(preview_image, preview_cnt, -1, (60, 169, 242), 1)
# horizontal
cv2.line(preview_image, (x + int(w / 2), y + int(h / 2)), (x + w, y + int(h / 2)), (255, 0, 0), 1)
# vertical
cv2.line(preview_image, (x + int(w / 2), y), (x + int(w / 2), y + int(h / 2)), (0, 0, 255), 1)
image = newimage
except Exception as exc:
exc_type, exc_obj, exc_tb = sys.exc_info()
fname = os.path.split(exc_tb.tb_frame.f_code.co_filename)[1]
print(exc_type, fname, exc_tb.tb_lineno)
print(type(self), exc)
return image, cnt_3d
def gcode(self, cnt_3d):
cnt = to_2D_contour(cnt_3d)
clearance = self.conf_file.get_float(key="clearance",
section=self.conf_section)
feed_rate = self.conf_file.get_float(key="feed_rate",
section=self.conf_section)
code = GCode()
code.feed_rate = feed_rate
code.rapid_rate = feed_rate * 2
code.clearance = clearance
if cnt_3d and len(cnt_3d) > 0:
# Determine the bounding rectangle
x, y, w, h = cv2.boundingRect(np.concatenate(cnt))
# scale contour from [micro m] to [mm]
scale_factor = 0.001
# transform all coordinates and generate gcode
# for this we must apply:
# - the scale factor
# - flip them upside down (gcode coordinate system vs. openCV coordinate system)
#
transform_y = lambda y_coord: (-(y_coord - y) + (y + h))
code.raise_mill()
code.feed_rapid({"x": 0, "y": 0})
code.start_spindle()
for c in cnt_3d:
i = 0
while i < len(c):
p = c[i]
grbl_x = '{:06.4f}'.format(p[0] * scale_factor)
grbl_y = '{:06.4f}'.format(
(transform_y(p[1])) * scale_factor)
grbl_z = '{:06.4f}'.format(p[2] * scale_factor)
position_save = {"x": grbl_x, "y": grbl_y}
position_carving = {"x": grbl_x, "y": grbl_y, "z": grbl_z}
if i == 0:
code.feed_rapid(position_save)
# move close to the surface
code.drop_mill()
# carve slowly into the workpiece until the bit has reached the final depth
code.feed_linear({"z": grbl_z})
else:
# move to the new x/y/z coordinate
code.feed_linear(position_carving)
i += 1
code.feed_linear({"z": 0})
code.raise_mill()
code.stop_spindle()
code.feed_rapid({"x": 0, "y": 0})
return code
def _process(self, image, cnt_3d):
if len(cnt_3d) > 0:
cnt = to_2D_contour(cnt_3d)
# Determine the bounding rectangle of all contours
x, y, w, h = cv2.boundingRect(np.concatenate(cnt))
image_height, image_width = image.shape[0], image.shape[1]
shift_x = int((x + (w / 2)) - image_width / 2)
shift_y = int((y + (h / 2)) - image_height / 2)
cnt_3d = [
np.subtract(c, [shift_x, shift_y, 0], dtype=np.int32)
for c in cnt_3d
]
cnt = to_2D_contour(cnt_3d)
newimage = np.zeros(image.shape, dtype="uint8")
newimage.fill(255)
cv2.drawContours(newimage, cnt, -1, (60, 169, 242), 1)
image = newimage
return image, cnt_3d
def process(self, image, cnt_3d):
if len(cnt_3d)>0:
cnt = to_2D_contour(cnt_3d)
# Determine the bounding rectangle of all contours
x, y, w, h = cv2.boundingRect(np.concatenate(cnt))
image_height, image_width = image.shape[0], image.shape[1]
# the offset to move the center of the contour to [0,0]
offset_x = x
offset_y = y
newimage = np.zeros(image.shape, dtype="uint8")
newimage.fill(255)
cnt_3d = [np.subtract(c, [offset_x, offset_y, 0], dtype=np.int32) for c in cnt_3d]
drawing_cnt = to_2D_contour(cnt_3d)
w2=image_width/2
h2=image_height/2
for c in drawing_cnt:
i = 0
while i < len(c):
p = c[i]
p[0] = p[0] + w2 - w/2
p[1] = p[1] + h2 - h/2
i+=1
x, y, w, h = cv2.boundingRect(np.concatenate(drawing_cnt))
# draw the width dimension
# horizontal
cv2.line(newimage, (x,y+h),(x+w,y+h), (255, 0, 0), 1)
# vertical
cv2.line(newimage, (x,y),(x,y+h), (0, 0, 255), 1)
cv2.drawContours(newimage, drawing_cnt, -1, (60,169,242), 1)
image = newimage
return image, cnt_3d
def process(self, image, cnt_3d):
if len(cnt_3d) > 0:
for c in cnt_3d:
for p in c:
p[2] = -self.depth_in_micro_m
cnt = to_2D_contour(cnt_3d)
newimage = np.zeros(image.shape, dtype="uint8")
newimage.fill(255)
# create a new cnt for the drawing on the image. The cnt should cover 4/5 of the overal image
#
image_height, image_width = image.shape[0], image.shape[1]
drawing_cnt = copy.deepcopy(cnt)
x, y, w, h = cv2.boundingRect(np.concatenate(drawing_cnt))
drawing_factor_w = (image_width / w) * 0.8
drawing_factor_h = (image_height / h) * 0.8
# ensure that the drawing fits into the preview image
drawing_factor = drawing_factor_w if drawing_factor_w < drawing_factor_h else drawing_factor_h
offset_x = (w / 2 + x) * drawing_factor
offset_y = (h / 2 + y) * drawing_factor
for c in drawing_cnt:
i = 0
while i < len(c):
p = c[i]
p[0] = (p[0] * drawing_factor) + (image_width /
2) - offset_x
p[1] = (p[1] * drawing_factor) + (image_height /
2) - offset_y
i += 1
cv2.drawContours(newimage, drawing_cnt, -1, (60, 169, 242), 1)
cv2.rectangle(newimage, (x, y), (x + w, y + h), (0, 255, 0), 1)
# draw the carving depth
cv2.putText(
newimage,
"Carving Depth {:.2f} mm".format(self.depth_in_micro_m / 1000),
(20, 50), cv2.FONT_HERSHEY_SIMPLEX, 0.8, (255, 0, 0), 4)
image = newimage
return image, cnt_3d
def process(self, image, cnt_3d):
if len(cnt_3d) > 0:
cnt = to_2D_contour(cnt_3d)
# Determine the bounding rectangle of all contours
x, y, w, h = cv2.boundingRect(np.concatenate(cnt))
image_height, image_width = image.shape[0], image.shape[1]
# the offset to move the center of the contour to [0,0]
offset_x = int(w / 2 + x)
offset_y = int(h / 2 + y)
cnt_3d = [
np.subtract(c, [offset_x, offset_y, 0], dtype=np.int32)
for c in cnt_3d
]
# shift the contour to the center. Only required for the drawing
#
w2 = int(image_width / 2) - offset_x
h2 = int(image_height / 2) - offset_y
drawing_cnt = [
np.subtract(c, [-w2, -h2], dtype=np.int32) for c in cnt
]
newimage = np.zeros(image.shape, dtype="uint8")
newimage.fill(255)
# draw the coordinate system of the centered drawing contour
x, y, w, h = cv2.boundingRect(np.concatenate(drawing_cnt))
cv2.drawContours(newimage, drawing_cnt, -1, (60, 169, 242), 1)
# horizontal
cv2.line(newimage, (x + int(w / 2), y + int(h / 2)),
(x + w, y + int(h / 2)), (255, 0, 0), 1)
# vertical
cv2.line(newimage, (x + int(w / 2), y),
(x + int(w / 2), y + int(h / 2)), (0, 0, 255), 1)
image = newimage
return image, cnt_3d
def _process(self, image, cnt_3d):
if len(cnt_3d) > 0:
for c in cnt_3d:
for p in c:
p[2] = -self.depth_in_micro_m
cnt = to_2D_contour(cnt_3d)
preview_image = np.zeros(image.shape, dtype="uint8")
preview_image.fill(255)
# create a new cnt for the drawing on the image. The cnt should cover 4/5 of the overall image
preview_cnt = contour_into_image(cnt, preview_image)
cv2.drawContours(preview_image, preview_cnt, -1, (60, 169, 242), 1)
# draw the carving depth
cv2.putText(
preview_image,
"Carving Depth {:.2f} mm".format(self.depth_in_micro_m / 1000),
(20, 50), cv2.FONT_HERSHEY_SIMPLEX, 0.8, (255, 0, 0), 4)
image = preview_image
return image, cnt_3d
def _process(self, image, cnt_3d):
if len(cnt_3d) > 0:
def gauss_kernel(kernlen=21, nsig=3):
x = np.linspace(-nsig, nsig, kernlen)
kern1d = np.diff(st.norm.cdf(x))
return kern1d / kern1d.sum()
# repeat and insert the first and last value in the array n-times
# (expand the array at the beginning and end). Required for convolve function.
#
add_padding = lambda array, n: np.insert(
np.insert(array, 0, np.full(n, array[0])), -1,
np.full(n, array[-1]))
box = gauss_kernel(self.window)
padding = int(self.window / 2)
smoothed_cnt = []
for c in cnt_3d:
x, y, z = c.T
if len(z) > 0:
if len(z) > self.window:
x_new = np.convolve(add_padding(x, padding),
box,
mode="valid")
y_new = np.convolve(add_padding(y, padding),
box,
mode="valid")
z_new = np.convolve(add_padding(z, padding),
box,
mode="valid")
# replace the starting points and end points of the CNC contour with the original points.
# We don't want modify the start / end of an contour. The reason for that is, that this is normaly
# cut-into movement into the stock without any tolerances. So - don't modify them.
#
# Replace the START Points with the original one
x_new[0:padding] = x[0:padding]
y_new[0:padding] = y[0:padding]
z_new[0:padding] = z[0:padding]
# Replace the END Points with the original one
x_new[-padding:] = x[-padding:]
y_new[-padding:] = y[-padding:]
z_new[-padding:] = z[-padding:]
smoothed_cnt.append(
np.asarray([[int(i[0]),
int(i[1]),
int(i[2])]
for i in zip(x_new, y_new, z_new)]))
else:
smoothed_cnt.append(c)
# generate a preview image
#
preview_image = np.zeros(image.shape, dtype="uint8")
preview_image.fill(255)
# generate a preview contour
#
preview_cnt = contour_into_image(to_2D_contour(smoothed_cnt),
preview_image)
# draw the centered contour
cv2.drawContours(preview_image, preview_cnt, -1, (60, 169, 242), 1)
image = preview_image
cnt_3d = smoothed_cnt
return image, cnt_3d
def process(self, image, cnt_3d):
if len(cnt_3d) > 0:
cnt = to_2D_contour(cnt_3d)
display_factor = 0.001 if self.display_unit == "mm" else 0.0001
# Determine the origin bounding rectangle
x, y, w, h = cv2.boundingRect(np.concatenate(cnt))
# Ensure that width of the contour is the same as the width_in_mm.
# Scale the contour to the required width.
scaled_factor = self.width_in_micro_m / w
scaled_cnt = [
np.multiply(c.astype(np.float),
[scaled_factor, scaled_factor, 1]).astype(np.int32)
for c in cnt_3d
]
# create a new cnt for the drawing on the image. The cnt should cover 4/5 of the overall image
#
height_in_micro_m = self.width_in_micro_m / w * h
image_height, image_width = image.shape[0], image.shape[1]
drawing_factor_w = (image_width / w) * 0.8
drawing_factor_h = (image_height / h) * 0.8
# ensure that the drawing fits into the preview image
drawing_factor = drawing_factor_w if drawing_factor_w < drawing_factor_h else drawing_factor_h
offset_x = (image_width / 2) - (w / 2 + x) * drawing_factor
offset_y = (image_height / 2) - (h / 2 + y) * drawing_factor
drawing_cnt = [
np.add(
np.multiply(c.astype(np.float),
[drawing_factor, drawing_factor]),
[offset_x, offset_y]).astype(np.int32) for c in cnt
]
# generate a preview image
#
newimage = np.zeros(image.shape, dtype="uint8")
newimage.fill(255)
# determine the drawing bounding box
x, y, w, h = cv2.boundingRect(np.concatenate(drawing_cnt))
# draw the centered contour
cv2.drawContours(newimage, drawing_cnt, -1, (60, 169, 242), 1)
cv2.rectangle(newimage, (x, y), (x + w, y + h), (0, 255, 0), 2)
# draw the width dimension
cv2.line(newimage, (x, y + int(h / 2)), (x + w, y + int(h / 2)),
(255, 0, 0), 2)
cv2.circle(newimage, (x, y + int(h / 2)), 15, (255, 0, 0), -1)
cv2.circle(newimage, (x + w, y + int(h / 2)), 15, (255, 0, 0), -1)
cv2.putText(
newimage,
"{:.1f} {}".format(self.width_in_micro_m * display_factor,
self.display_unit),
(x + 20, y + int(h / 2) - 30), cv2.FONT_HERSHEY_SIMPLEX, 1.65,
(255, 0, 0), 4)
# draw the height dimension
cv2.line(newimage, (x + int(w / 2), y), (x + int(w / 2), y + h),
(255, 0, 0), 2)
cv2.circle(newimage, (x + int(w / 2), y), 15, (255, 0, 0), -1)
cv2.circle(newimage, (x + int(w / 2), y + h), 15, (255, 0, 0), -1)
cv2.putText(
newimage,
"{:.1f} {}".format(height_in_micro_m * display_factor,
self.display_unit),
(x + int(w / 2) + 20, y + 50), cv2.FONT_HERSHEY_SIMPLEX, 1.65,
(255, 0, 0), 4)
image = newimage
cnt_3d = scaled_cnt
return image, cnt_3d
def _process(self, image, cnt):
try:
PADDING = 2
# add padding to the image to avoid that we run out of index in the np.amax calculation
image = cv2.copyMakeBorder(image, PADDING, PADDING, PADDING,
PADDING, cv2.BORDER_CONSTANT, None,
[255, 255, 255])
image_height, image_width = image.shape[0], image.shape[1]
# generate the skeleton of the "BLACK" area and returns an image in which the
# "white" part is the skeleton.
#
image_skeleton = image.copy(
) # deepcopy to protect the original image
image_skeleton = cv2.threshold(image_skeleton, 127, 255,
cv2.THRESH_BINARY_INV)[1]
image_skeleton, _, _ = cv2.split(image_skeleton)
#image_skeleton = 255 - cv2.ximgproc.thinning(image_skeleton, cv2.ximgproc.THINNING_GUOHALL)
image_skeleton = cv2.ximgproc.thinning(
image_skeleton, cv2.ximgproc.THINNING_ZHANGSUEN)
# calculate the watershed distance. This is a distance calculation of the original image
#
image_watershed = 255 - cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
image_watershed = cv2.distanceTransform(image_watershed,
cv2.DIST_L2, 5)
image_watershed = cv2.convertScaleAbs(image_watershed)
image_watershed = cv2.normalize(image_watershed, None, 255, 0,
cv2.NORM_MINMAX, cv2.CV_8UC1)
# calculate the mask and use just the pixel and the intensity for the carving depth
# now we have just the skeleton image with the distance information (carving depth)
image_carvingpath = cv2.bitwise_and(image_watershed,
image_skeleton)
# CHAIN_APPROX_NONE is important to get each and every pixel in the contour and not the reduced one.
#
cnt, hierarchy = cv2.findContours(image_skeleton, cv2.RETR_TREE,
cv2.CHAIN_APPROX_NONE)
cnt = normalize_contour(cnt)
x, y, w, h = cv2.boundingRect(np.concatenate(cnt))
preview_image = np.zeros(image.shape, dtype="uint8")
preview_image.fill(255)
generated_cnt = ensure_3D_contour(cnt)
dots_in_width = int(self.width_in_micro_m /
min(self.cutter_bit_diameter_in_micro_m,
self.max_diameter_in_micro_m))
max_diameter_in_pixel = (image_width / dots_in_width)
carving_depth = lambda gray: -(
self.max_diameter_in_micro_m / 255 * (gray)) / (math.tan(
math.radians(self.cutter_bit_angle / 2)) * 2)
circle_radius = lambda gray: int(((max_diameter_in_pixel / 255) *
(gray)) / 2)
for c in generated_cnt:
i = 0
while i < len(c):
p = c[i]
row = p[1]
col = p[0]
neighbours = image_carvingpath[row - 2:row + 2,
col - 2:col + 2]
max = np.amax(neighbours)
p[2] = carving_depth(max)
cv2.circle(preview_image, (col, row), circle_radius(max),
(60, 169, 242), -1)
i = i + 1
cv2.drawContours(preview_image, to_2D_contour(generated_cnt), -1,
(0, 49, 252), 1)
display_factor = 0.001 if self.display_unit == "mm" else 0.0001
# draw the width dimension
cv2.line(preview_image, (x, y + int(h / 2)),
(x + w, y + int(h / 2)), (255, 0, 0), 1)
cv2.circle(preview_image, (x, y + int(h / 2)), 5, (255, 0, 0), -1)
cv2.circle(preview_image, (x + w, y + int(h / 2)), 5, (255, 0, 0),
-1)
cv2.putText(
preview_image,
"{:.1f} {}".format(self.width_in_micro_m * display_factor,
self.display_unit),
(x + 20, y + int(h / 2) - 30), cv2.FONT_HERSHEY_SIMPLEX, 1.65,
(255, 0, 0), 4)
# draw the height dimension
height_in_micro_m = self.width_in_micro_m / w * h
cv2.line(preview_image, (x + int(w / 2), y),
(x + int(w / 2), y + h), (255, 0, 0), 1)
cv2.circle(preview_image, (x + int(w / 2), y), 5, (255, 0, 0), -1)
cv2.circle(preview_image, (x + int(w / 2), y + h), 5, (255, 0, 0),
-1)
cv2.putText(
preview_image,
"{:.1f} {}".format(height_in_micro_m * display_factor,
self.display_unit),
(x + int(w / 2) + 20, y + 50), cv2.FONT_HERSHEY_SIMPLEX, 1.65,
(255, 0, 0), 4)
# Ensure that width of the contour is the same as the width_in_mm.
# Scale the contour to the required width.
scale_factor = self.width_in_micro_m / w
generated_cnt = [
np.multiply(c.astype(np.float),
[scale_factor, scale_factor, 1]).astype(np.int32)
for c in generated_cnt
]
preview_image = preview_image[PADDING:-PADDING, PADDING:-PADDING]
return preview_image, generated_cnt
except Exception as exc:
exc_type, exc_obj, exc_tb = sys.exc_info()
fname = os.path.split(exc_tb.tb_frame.f_code.co_filename)[1]
print(exc_type, fname, exc_tb.tb_lineno)
print(type(self), exc)
def _process(self, image, cnt_3d):
if len(cnt_3d) > 0:
cnt = to_2D_contour(cnt_3d)
display_factor = 0.001 if self.display_unit == "mm" else 0.0001
# Determine the origin bounding rectangle
x, y, w, h = cv2.boundingRect(np.concatenate(cnt))
# Ensure that width of the contour is the same as the width_in_mm.
# Scale the contour to the required width.
scaled_factor = self.width_in_micro_m / w
scaled_cnt = [
np.multiply(c.astype(np.float),
[scaled_factor, scaled_factor, 1]).astype(np.int32)
for c in cnt_3d
]
# generate a preview image
#
preview_image = np.zeros(image.shape, dtype="uint8")
preview_image.fill(255)
# generate a preview contour
#
preview_cnt = contour_into_image(to_2D_contour(scaled_cnt),
preview_image)
# determine the drawing bounding box
x, y, w, h = cv2.boundingRect(np.concatenate(preview_cnt))
# draw the centered contour
cv2.drawContours(preview_image, preview_cnt, -1, (60, 169, 242), 1)
# draw the width dimension
cv2.line(preview_image, (x, y + int(h / 2)),
(x + w, y + int(h / 2)), (255, 0, 0), 1)
cv2.circle(preview_image, (x, y + int(h / 2)), 5, (255, 0, 0), -1)
cv2.circle(preview_image, (x + w, y + int(h / 2)), 5, (255, 0, 0),
-1)
cv2.putText(
preview_image,
"{:.1f} {}".format(self.width_in_micro_m * display_factor,
self.display_unit),
(x + 20, y + int(h / 2) - 30), cv2.FONT_HERSHEY_SIMPLEX, 1.65,
(255, 0, 0), 4)
# draw the height dimension
height_in_micro_m = self.width_in_micro_m / w * h
cv2.line(preview_image, (x + int(w / 2), y),
(x + int(w / 2), y + h), (255, 0, 0), 1)
cv2.circle(preview_image, (x + int(w / 2), y), 5, (255, 0, 0), -1)
cv2.circle(preview_image, (x + int(w / 2), y + h), 5, (255, 0, 0),
-1)
cv2.putText(
preview_image,
"{:.1f} {}".format(height_in_micro_m * display_factor,
self.display_unit),
(x + int(w / 2) + 20, y + 50), cv2.FONT_HERSHEY_SIMPLEX, 1.65,
(255, 0, 0), 4)
image = preview_image
cnt_3d = scaled_cnt
return image, cnt_3d