|
|
|
|
@ -69,22 +69,31 @@ FLAG_gauss = 0
|
|
|
|
|
FLAG_uniform = 1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def process_image_part(data_part, kernel):
|
|
|
|
|
def process_image_part(data_part, kernel, padding):
|
|
|
|
|
y_part_size, x_part_size, _ = data_part.shape
|
|
|
|
|
data_part_new = np.zeros_like(data_part)
|
|
|
|
|
for i in range((kernel.shape[0] - 1) // 2, y_part_size - (kernel.shape[0] - 1) // 2):
|
|
|
|
|
for j in range((kernel.shape[1] - 1) // 2, x_part_size - (kernel.shape[1] - 1) // 2):
|
|
|
|
|
|
|
|
|
|
data_part_new = np.zeros((data_part.shape[0] - padding[0], data_part.shape[1] - padding[1], 3))
|
|
|
|
|
|
|
|
|
|
pad_y, pad_x = padding
|
|
|
|
|
for i in range(pad_y, y_part_size - pad_y):
|
|
|
|
|
for j in range(pad_x, x_part_size - pad_x):
|
|
|
|
|
for k in range(3):
|
|
|
|
|
new_value = 0.0
|
|
|
|
|
for ii in range(kernel.shape[0]):
|
|
|
|
|
for jj in range(kernel.shape[1]):
|
|
|
|
|
iii = ii - (kernel.shape[0] - 1) // 2
|
|
|
|
|
jjj = jj - (kernel.shape[1] - 1) // 2
|
|
|
|
|
iii = ii - pad_y
|
|
|
|
|
jjj = jj - pad_x
|
|
|
|
|
new_value += kernel[ii, jj] * data_part[i + iii, j + jjj, k]
|
|
|
|
|
data_part_new[i, j, k] = new_value
|
|
|
|
|
data_part_new[i - pad_y, j - pad_x, k] = new_value
|
|
|
|
|
return data_part_new
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def split_array(arr, n, overlap):
|
|
|
|
|
sub_array_length = (len(arr) + (n - 1) * overlap) // n
|
|
|
|
|
sub_arrays = [arr[i * (sub_array_length - overlap): i * (sub_array_length - overlap) + sub_array_length] for i in range(n)]
|
|
|
|
|
return np.array(sub_arrays)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def process_image(img, func=0, mode=0):
|
|
|
|
|
if isinstance(img, str):
|
|
|
|
|
img = Image.open(img)
|
|
|
|
|
@ -95,15 +104,19 @@ def process_image(img, func=0, mode=0):
|
|
|
|
|
data = np.asarray(img, dtype=np.float64) / 255.0
|
|
|
|
|
|
|
|
|
|
if func == 1:
|
|
|
|
|
kernel = uniform_kernel(3)
|
|
|
|
|
kernel = uniform_kernel(7)
|
|
|
|
|
else:
|
|
|
|
|
kernel = gauss_kernel(3)
|
|
|
|
|
|
|
|
|
|
padding = [(kernel.shape[0] // 2), kernel.shape[1] // 2]
|
|
|
|
|
|
|
|
|
|
if mode == 1: # Multithreading mode
|
|
|
|
|
num_threads = 16
|
|
|
|
|
data_parts = np.array_split(data, num_threads, axis=0)
|
|
|
|
|
|
|
|
|
|
data_parts = split_array(data, num_threads, padding[0])
|
|
|
|
|
|
|
|
|
|
with multiprocessing.Pool(num_threads) as pool:
|
|
|
|
|
data_new_parts = pool.starmap(process_image_part, zip(data_parts, [kernel]*num_threads))
|
|
|
|
|
data_new_parts = pool.starmap(process_image_part, zip(data_parts, [kernel]*num_threads, [padding]*num_threads))
|
|
|
|
|
data_new = np.concatenate(data_new_parts, axis=0)
|
|
|
|
|
|
|
|
|
|
elif mode == 2: # MPI parallel mode
|
|
|
|
|
@ -111,7 +124,7 @@ def process_image(img, func=0, mode=0):
|
|
|
|
|
rank = comm.Get_rank()
|
|
|
|
|
size = comm.Get_size()
|
|
|
|
|
data_part = np.array_split(data, size, axis=0)[rank]
|
|
|
|
|
data_new_part = process_image_part(data_part, kernel)
|
|
|
|
|
data_new_part = process_image_part(data_part, kernel, padding)
|
|
|
|
|
data_new_parts = comm.gather(data_new_part, root=0)
|
|
|
|
|
if rank == 0:
|
|
|
|
|
data_new = np.concatenate(data_new_parts, axis=0)
|
|
|
|
|
@ -120,7 +133,7 @@ def process_image(img, func=0, mode=0):
|
|
|
|
|
data_new = comm.bcast(data_new, root=0)
|
|
|
|
|
|
|
|
|
|
else: # Default mode
|
|
|
|
|
data_new = process_image_part(data, kernel)
|
|
|
|
|
data_new = process_image_part(data, kernel, padding)
|
|
|
|
|
|
|
|
|
|
data_new = data_new * 255.0
|
|
|
|
|
data_new = np.uint8(data_new)
|
|
|
|
|
@ -128,6 +141,7 @@ def process_image(img, func=0, mode=0):
|
|
|
|
|
return Image.fromarray(data_new, mode="RGB")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
if __name__ == '__main__':
|
|
|
|
|
print(monte_carlo_methode(1000, FLAG_default))
|
|
|
|
|
print(monte_carlo_methode(1000, FLAG_threaded))
|
|
|
|
|
|
