I have been doing some work deconvoluting images with 2D Scipy FFTs. However, Matplotlib for no apparent reason is inverting the color scheme of the generated IFFT array, even though the RGB values are correct.
import numpy as np
from scipy import fftpack
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
image = mpimg.imread("C:/Users/-----/Desktop/image.jpg")
freq = fftpack.fft2(image)
IFFT = fftpack.ifft2(freq)
IFFT = IFFT.astype('float32')
plt.figure(1)
plt.imshow(image)
plt.figure(2)
plt.imshow(IFFT)
plt.show()
The IFFT array and image array are equal according to numpy.array_equal, and yet the colormap of the second figure is always inverted. See the attached images. The arrays are literally identical and no other colormap is specified, and yet I am forced to manual invert everything using something like this:
for i in range(0, freq.shape[1]):
for j in range (0, freq.shape[0]):
for k in range(0,3):
freq[j,i,k] = 255-freq[j,i,k]
I wonder if the astype conversion to float32 (or earlier uint32) might be changing something, but since arrays are identical, I have no idea.
Any ideas? I'd also like to figure out how to invert the entire cmap if that would be a more efficient alternative to manually subtracting 255 from every entry in the array.
