Spaces:
Running
Running
File size: 1,848 Bytes
911c613 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 |
# ------------------------------------------------------------ #
#
# file : utils/learning/patch/reconstruction.py
# author : CM
# Function to reconstruct image from patch
#
# ------------------------------------------------------------
import numpy as np
# ----- Image Reconstruction -----
# Recreate the image from patchs
def fullPatchsToImage(image,patchs):
i = 0
for x in range(0,image.shape[0], patchs.shape[1]):
for y in range(0, image.shape[1], patchs.shape[2]):
for z in range(0,image.shape[2], patchs.shape[3]):
image[x:x+patchs.shape[1],y:y+patchs.shape[2],z:z+patchs.shape[3]] = patchs[i,:,:,:,0]
i = i+1
return image
def fullPatchsPlusToImage(image,patchs, dx, dy, dz):
div = np.zeros(image.shape)
one = np.ones((patchs.shape[1],patchs.shape[2],patchs.shape[3]))
i = 0
for x in range(0,image.shape[0]-dx, dx):
for y in range(0, image.shape[1]-dy, dy):
for z in range(0,image.shape[2]-dz, dz):
div[x:x+patchs.shape[1],y:y+patchs.shape[2],z:z+patchs.shape[3]] += one
image[x:x+patchs.shape[1],y:y+patchs.shape[2],z:z+patchs.shape[3]] = patchs[i,:,:,:,0]
i = i+1
image = image/div
return image
def dolzReconstruction(image,patchs):
output = np.copy(image)
count = 0
print("image shape", image.shape)
print("patchs shape", patchs.shape)
# todo : change 16 with patch shape
for x in range(0,image.shape[0], 16):
for y in range(0, image.shape[1], 16):
for z in range(0,image.shape[2], 16):
patch = np.argmax(patchs[count], axis=1)
patch = patch.reshape(16, 16, 16)
output[x:x+patch.shape[0],y:y+patch.shape[1],z:z+patch.shape[2]] = patch
count += 1
return output |