CH02_SEC06_1_2DFFT.ipynb

Kernel: Python 3

from matplotlib.image import imread
import numpy as np
import matplotlib.pyplot as plt
import os
plt.rcParams['figure.figsize'] = [12, 8]
plt.rcParams.update({'font.size': 18})

A = imread(os.path.join('..','DATA','dog.jpg'))
B = np.mean(A, -1); # Convert RGB to grayscale

fig,axs = plt.subplots(1,3)

# Plot image
img = axs[0].imshow(B)
img.set_cmap('gray')
axs[0].axis('off')

# Compute row-wise FFT

Cshift = np.zeros_like(B,dtype='complex_')
C = np.zeros_like(B,dtype='complex_')

for j in range(B.shape[0]):
    Cshift[j,:] = np.fft.fftshift(np.fft.fft(B[j,:]))
    C[j,:] = np.fft.fft(B[j,:])
    
img = axs[1].imshow(np.log(np.abs(Cshift)))
img.set_cmap('gray')
axs[1].axis('off')

# Compute column-wise FFT

D = np.zeros_like(C)
for j in range(C.shape[1]):
    D[:,j] = np.fft.fft(C[:,j])

img = axs[2].imshow(np.fft.fftshift(np.log(np.abs(D))))
img.set_cmap('gray')
axs[2].axis('off')

plt.show()

# Much more efficient to use fft2
D = np.fft.fft2(B)