Files
scikit-image/skimage/transform/rescale.py
T

47 lines
1.2 KiB
Python

# TODO : Doc, Tests, PEP8 check
import numpy as np
def downsample(image, factors, method='sum'):
is0 = image.shape[0]
is1 = image.shape[1]
f0 = factors[0]
f1 = factors[1]
if (f0 - int(f0) != 0) or (f1 - int(f1) != 0):
print "Use resample() for non-integer downsampling"
return
cropped = image[:is0 - (is0 % f0), :is1 - (is1 % f1)]
out = np.zeros((cropped.shape[0] / f0, cropped.shape[1] / f1))
for i in range(cropped.shape[0]):
for j in range(cropped.shape[1]):
out[int(i / f0)][int(j / f1)] += cropped[i][j]
if method == 'sum':
return out
else:
return out / float(f0 * f1)
def upsample(image, factors, method='divide'):
is0 = image.shape[0]
is1 = image.shape[1]
f0 = factors[0]
f1 = factors[1]
if (f0 - int(f0) != 0) or (f1 - int(f1) != 0):
print "Use resample() for non-integer upsampling"
return
out = np.zeros((f0 * image.shape[0], f1 * image.shape[1]))
for i in range(out.shape[0]):
for j in range(out.shape[1]):
out[i][j] = (image[i / f0][j / f1])
if method == 'divide':
return out / float(f0 * f1)
else:
return out