def __applymeanthreshold(self, blob_img):
size = blob_img.shape
height = size[0]
width = size[1]
mask = np.zeros((height, width), dtype=np.uint8)
smooth_projection_x = self.smooth_projection[0]
mean_x = int(self.mean[0] / 4)
smooth_projection_y = self.smooth_projection[1]
mean_y = int(self.mean[1] / 3)
# Mask in X
for ii in xrange(0, height - 1, 1):
if (smooth_projection_x[ii] >= mean_x):
mask[ii, :] = 255
# Mask in Y
for jj in xrange(0, width - 1, 1):
if (smooth_projection_y[jj] >= mean_y):
mask[:, jj] = 255
# By now only mask in y projection
self.mask = cv2.bitwise_and(blob_img, blob_img, mask=mask)
def __applymeanthreshold(self, blob_img):
size = blob_img.shape
height = size[0]
width = size[1]
mask = np.zeros((height, width), dtype=np.uint8)
smooth_projection_x = self.smooth_projection[0]
mean_x = int((self.mean[0]*2) / 3)
smooth_projection_y = self.smooth_projection[1]
mean_y = int((self.mean[1]*2) / 3)
# Mask in X
for ii in xrange(0, height - 1, 1):
if(smooth_projection_x[ii] >= mean_x):
mask[ii, :] = 255
# Mask in Y
for jj in xrange(0, width - 1, 1):
if(smooth_projection_y[jj] >= mean_y):
mask[:, jj] = 255
# By now only mask in y projection
self.mask = cv2.bitwise_and(blob_img, blob_img, mask=mask)
def drawmask(self, frame):
# Create colored mask for visualization
x, y, w, h = self.bound_rect
mask_color = np.zeros((h, w, 3), dtype=np.uint8)
mask_color[:, :, 1] = 255 # Assign blue color to created colored mask
mask_color = cv2.bitwise_and(mask_color, mask_color, mask=self.mask)
frame[y:y + h, x:x + w] = cv2.add(frame[y:y + h, x:x + w], mask_color)
def drawmask(self, frame):
# Create colored mask for visualization
x, y, w, h = self.bound_rect
mask_color = np.zeros((h, w, 3), dtype=np.uint8)
mask_color[:, :, 1] = 255 # Assign blue color to created colored mask
mask_color = cv2.bitwise_and(mask_color, mask_color, mask=self.mask)
frame[y:y + h, x:x + w] = cv2.add(frame[y:y + h, x:x + w], mask_color)
def globalmasktosubjects(total_masks, bg_models, cameras, frames):
debug_flag = 0
if debug_flag:
print(">> Detecciones")
total_subjs = []
win_width = bg_models[0].win_width
win_height = bg_models[0].win_height
for ii in range(len(total_masks)):
subjs = []
# OpenCV 2.4.8
contours, hierarchy = cv2.findContours(
total_masks[ii], cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
"""
# OpenCV 3.0.0
_, contours, hierarchy = cv2.findContours(
total_masks[ii], cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
"""
res = cv2.bitwise_and(frames[ii], frames[ii], mask = total_masks[ii])
for cont in contours:
x, y, w, h = cv2.boundingRect(cont)
box = [x, y, w, h]
if w >= (win_width / 2) and h >= (win_height / 2):
ellipse = cv2.fitEllipse(cont)
rot_box = cv2.minAreaRect(cont)
circle = cv2.minEnclosingCircle(cont)
subj = subject.Subject()
subj.setdefault(
total_masks[ii][y:y + h, x:x + w],
box, rot_box, ellipse, circle, cameras[ii], cont,
res[y:y + h, x:x + w])
subjs.append(subj)
if debug_flag:
print(box)
total_subjs.append(subjs)
return total_subjs
def globalmasktosubjects(total_masks, bg_models, cameras, frames):
debug_flag = 0
if debug_flag:
print(">> Detecciones")
total_subjs = []
win_width = bg_models[0].win_width
win_height = bg_models[0].win_height
for ii in range(len(total_masks)):
subjs = []
# OpenCV 2.4.8
contours, hierarchy = cv2.findContours(total_masks[ii],
cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_NONE)
"""
# OpenCV 3.0.0
_, contours, hierarchy = cv2.findContours(
total_masks[ii], cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
"""
res = cv2.bitwise_and(frames[ii], frames[ii], mask=total_masks[ii])
for cont in contours:
x, y, w, h = cv2.boundingRect(cont)
box = [x, y, w, h]
if w >= (win_width / 2) and h >= (win_height / 2):
ellipse = cv2.fitEllipse(cont)
rot_box = cv2.minAreaRect(cont)
circle = cv2.minEnclosingCircle(cont)
subj = subject.Subject()
subj.setdefault(total_masks[ii][y:y + h, x:x + w], box,
rot_box, ellipse, circle, cameras[ii], cont,
res[y:y + h, x:x + w])
subjs.append(subj)
if debug_flag:
print(box)
total_subjs.append(subjs)
return total_subjs
