Skip to content

Instantly share code, notes, and snippets.

What would you like to do?
#!/bin/env python3
from import imread
import numpy as np
import matplotlib.pyplot as plt
import as cm
# Load the picture
img = imread('./lenin.jpg', flatten=True, mode=None)
# decompose the picture matrix
U, s, V = np.linalg.svd(img, full_matrices=False)
rank = np.linalg.matrix_rank(img, tol=None)
# Take first 25 singular values (25 biggest values), you should play with it and assess quality
approx_rank = 25
compression = approx_rank * (img.shape[0] + img.shape[1]) / (img.shape[0] * img.shape[1])
real_img =,, V))
S = np.diag(s)
# restore pictures, but with only approx_rank singular values
approx_img =
U[:, 0:approx_rank],
S[0:approx_rank, 0:approx_rank],
V[0:approx_rank, :]))
plt.subplot(1, 2, 1)
plt.imshow(real_img, cmap=cm.Greys_r)
plt.xlabel('Real image rank: ' + str(rank))
plt.subplot(1, 2, 2)
plt.imshow(approx_img, cmap=cm.Greys_r)
plt.xlabel('Approx rank: ' + str(approx_rank) + ' compression: {:2.4f}'.format(compression))
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
You can’t perform that action at this time.