numpy 2D数组的怪异插值

Question

I've got a bunch of data sets in the form of 131 by 20 2D numpy arrays and I'm trying to interpolate those to square 131x131 ones. I've seen others' code many times now but I'm still getting some odd results. Here is my interpolation code:

import numpy as np
import matplotlib.pyplot as plt
from scipy import interpolate

# bunch of code

# radarray.shape = (131, 20)

T0 = np.arange(0, radarray.shape[1], 1)
Z0 = np.arange(0, radarray.shape[0], 1)

t, z = np.meshgrid(T0, Z0)

f = interpolate.interp2d(t, z, radarray, kind='quintic')

T = np.linspace(0, radarray.shape[1], 131)
Z = np.arange(0, radarray.shape[0], 1)
radinterp = f(T, Z)
plt.imshow(radinterp, aspect='auto', interpolation='none')
plt.gca().invert_yaxis()
plt.savefig('interp')

This is what the raw data from radarray looks like:

And this is the result:

I've tried switching the interpolation methods and the radarray.shape indices but everything results in just a green field or these weird vaguely sine bumps that have nothing to do with my data. Where did I screw up?

Answer 1

Why don't you use skimage's resize ?

from skimage.transform import resize

radinterp = resize(radarray, (131, 131), anti_aliasing=True)

Should do the trick for you.