def get_basic_section(start, classified_pixels, start_back=None):
# 'gradient' vector
delta = np.array([0, 0])
last_gradients = CircularList()
section = []
section.append(start)
x, y = start
next, next_value = get_max_neighbor(classified_pixels, x, y, start_back)
delta = np.subtract(next, start)
while next_value == 2: # edge pixel
last_gradients.insert((delta[0], delta[1]))
section.append(next)
next = np.add(next, delta)
next_value = classified_pixels[next[0], next[1]]
if next_value < 2: # blank pixel or miscellaneous pixel
last = section[-1] # get last element added in section
x, y = last
back = np.subtract(last, delta)
# get max value in the neighborhood, unless the 'back'
next, next_value = get_max_neighbor(classified_pixels, x, y, back)
delta = np.subtract(next, last)
last_gradients.insert((delta[0], delta[1]))
section.append(next)
last_element = next
return section, last_gradients, last_element
def get_crossing_section_direction(classified_pixels, crossing_pixel,
last_gradients, section):
# counter gradients frequency
cnt_gradient = Counter(last_gradients.get_list())
# count in list
grads = cnt_gradient.most_common()
crossing_section_direction = []
next = crossing_pixel
next_value = classified_pixels[next[0], next[1]]
# back is a edge pixel
back = section[-2][0], section[-2][1]
# avoid local minima
iterations = 0
loop_grads = CircularList(3)
excluded_grad = None
while next_value != 2: # edge pixel
aux_value = 0
i = 0
if iterations == 3:
list_loop_grads = loop_grads.get_list()
excluded_grad = list_loop_grads[1]
crossing_section_direction[:] = []
iterations = 0
# blank pixel or miscellaneous and i < len
while aux_value < 2 and i < len(grads):
if grads[i][0] == excluded_grad:
continue
delta = grads[i][0]
aux = np.add(next, delta)
if aux[0] == back[0] and aux[1] == back[
1]: # back[0] >= 0 and back[1] >= 0 and
aux_value = 0
else:
aux_value = classified_pixels[aux[0], aux[1]]
i += 1
if aux_value < 2 and i == len(grads):
delta = get_direction(classified_pixels, back, next, grads,
excluded_grad)
loop_grads.insert(delta)
back = next[0], next[1]
next = np.add(next, delta)
next_value = classified_pixels[next[0], next[1]]
else:
loop_grads.insert(delta)
back = next[0], next[1]
next = aux
next_value = aux_value
crossing_section_direction.append(next)
iterations += 1
return crossing_section_direction
