Some warning and error fix for chapter 01 02 03

With python 3.9 and up-to-date libraries
Chapter 01
Perspective Projection and Homography
matplotlib 3.4 savefig bbox_in, pad_in is deprecated, updated to bbox_inches and pad_inches
plt.savefig('images/homography_out.png', bbox_inches='tight', pad_inches=0)
Pencil Sketches from images
plt.savefig(img_file.split('.')[0] + '_sketches_all.png', bbox_inches='tight')
Image need using tuple() instead array warning

#output = img - anisotropic_diffusion(img, niter=niter, kappa=kappa, gamma=gamma, voxelspacing=None, option=1) # change to
output = img - anisotropic_diffusion((img), niter=niter, kappa=kappa, gamma=gamma, voxelspacing=None, option=1)

Chapter 02
Edge Detection with Canny, LOG / Zero-Crossing and Wavelets

from skimage.io import imread # use io package as misc imread deprecrated
#img = rgb2gray(misc.imread('images/tiger.png')) # change to
img = rgb2gray(imread('images/tiger.png'))

Edge Detection with Anisotropic Diffusion
#diff_out = anisotropic_diffusion(img, niter=50, kappa=20, option=1)
diff_out = anisotropic_diffusion((img), niter=50, kappa=20, option=1)
Image Denoising with Denoising Autoencoder
import torchvision, matplotlib, sklearn, numpy as np, torch# added torch
Improving Image Contrast

#hist, bins = np.histogram(img[...,i].flatten(),256,[0,256], normed=True) # Changed to 
hist, bins = np.histogram(img[...,i].flatten(),256,[0,256], density=True)  
#plt.savefig('images/hist_out.png', bbox_in='tight', pad_in=0) # Changed to 
plt.savefig('images/hist_out.png', bbox_inches='tight', pad_inches=0)

Image Denoising with Anisotropic Diffusion

#diff_out = anisotropic_diffusion(noisy, niter=20, kappa=20, option=1) 
diff_out = anisotropic_diffusion((noisy), niter=20, kappa=20, option=1)
#diff_out = anisotropic_diffusion(noisy, niter=50, kappa=100, option=2)
diff_out = anisotropic_diffusion((noisy), niter=50, kappa=100, option=2)

Chapter 03
Wiener Filter
this import only support until < 0.16.2
from skimage.measure import compare_psnr
need update to
from skimage.metrics import peak_signal_noise_ratio as compare_psnr

#ax = fig.gca(projection='3d')
ax = fig.add_subplot(projection='3d')

Some warning updated on CLA section

from skimage.color import rgb2gray,rgba2rgb
im = rgb2gray(rgba2rgb(imread('images/book.png'))) # street

Noisy Image Restoration with Markov Random Field

#% matplotlib inline
%matplotlib inline

Image Completion with Inpainting (using Deep learning - pre-trained torch CompletionNet model)
Only with python 3.7 torch==0.4.1 torchvision==0.2.0

#from torch.legacy import nn
#from torch.legacy.nn.Sequential import Sequential
from torch import nn
from torch.nn import Sequential

Image Restoration with Dictionary Learning
Online Dictionary Learning

#lena = rgb2gray(imread('images/lena.png'))
lena = rgb2gray(rgba2rgb(imread('images/lena.png')))

Image Compression with Wavelets

#imw = soft_threshold(imw, 12)
imw = soft_threshold((imw), 12)

Inspectez les notebooks ou les exemples de Chapter 01, Chapter 02 et Chapter 03, puis exécutez-les avec Python 3.9 et des bibliothèques à jour afin de reproduire les avertissements et les erreurs indiqués. Appliquez les mises à jour de compatibilité documentées et vérifiez que les exemples de traitement d’images concernés s’exécutent sans avertissements de dépréciation ni échecs.