def gauss(array, mode, sigma):
"""
Фильтр Гаусса
array: двумерный массив
mode: тип экстраполяции
sigma: радиус размытия
result: двумерный массив
"""
kernel = gauss_filter1d(sigma)
tmp = convolve1d(array, kernel, mode, axis = 1)
result = convolve1d(tmp, kernel, mode, axis = 0)
return result
def test_convolve1d_rep_y(self):
array = [[1, 2, 3, 4, 5], [1, 2, 3, 4, 5], [1, 2, 3, 4, 5]]
kernel = [1, 2, 3, 4]
output = tolist(
filters.convolve1d(array, kernel, axis=0, mode='nearest'))
result = conv.convolve1d(array, kernel, mode='rep', axis=0)
self.assertEqual(output, result)
def test_convolve1d_odd_y(self):
array = [[1, 2, 3], [1, 2, 3], [1, 2, 3]]
kernel = [1, 2, 3]
output = tolist(
filters.convolve1d(array, kernel, axis=0, mode='reflect'))
result = conv.convolve1d(array, kernel, mode='even', axis=0)
self.assertEqual(output, result)
def test_convolve1d_1d_odd2(self):
kernel_odd_size = [1, 2, 3, 4]
output = tolist(
filters.convolve1d(self.array1d, kernel_odd_size, mode='reflect'))
result = conv.convolve1d(self.array1d,
kernel_odd_size,
mode='even',
axis=-1)
self.assertEqual(output, result)
def test_convolve1d_1d_rep_1(self):
kernel_even_size = [1, 2, 3]
output = tolist(
filters.convolve1d(self.array1d, kernel_even_size, mode='nearest'))
result = conv.convolve1d(self.array1d,
kernel_even_size,
mode='rep',
axis=-1)
self.assertEqual(output, result)
def sobel(array, mode, direction):
"""
Фильтр собеля,применямый к двумерному массиву array.
Алгоритм реализуется применением одномерный сверток по строкам и по столбцам.
array: двумерный массив, для которого вычисляются градиенты
mode: тип продолжения
rep - дублирование граничных пикселей
odd - четное продолжение
even - нечетное продолжение
direction:
x - горизонтальные контуры
y - вертикальные контуры
result: двумерный массив
"""
if direction == "x":
tmp_by_rows = convolve1d(array, [1, 0, -1], mode=mode, axis=1)
return convolve1d(tmp_by_rows, [1, 2, 1], mode=mode, axis=0)
elif direction == "y":
tmp_by_rows = convolve1d(array, [1, 2, 1], mode=mode, axis=1)
return convolve1d(tmp_by_rows, [1, 0, -1], mode=mode, axis=0)
else:
raise ValueError("Unsupported direction : " + direction)
def _proccess_y(array, mode, sigma):
derivative_kernel = _gauss_derivative_filter(sigma)
gauss_kernel = gauss_filter1d(sigma)
tmp = convolve1d(array, derivative_kernel, mode, axis = 0)
return convolve1d(tmp, gauss_kernel, mode, axis = 1)
def test_convolve1d_1d_even(self):
array = [1, 2, 3, 4]
kernel = [1, 2, 3, 4]
output = [14, 20, 29, 37]
result = conv.convolve1d(array, kernel, mode='odd', axis=-1)
self.assertEqual(output, result)