Try using Pillow on iPython (Part 3)

Pillow work flow (3)

Load the image with Image.open
Convert image to grayscale
Gonyo Gonyo the image
Save the image edited with Image.save

`python`


%pylab inline
from PIL import Image,ImageDraw,ImageFont
img = Image.open('work-image/lena.jpg')
pl_img = np.array(img) ; plt.imshow( pl_img ) #display

`python`


#Convert to grayscale
gray_img = img.convert("L")

#Not the intended display
pl_img = np.array(gray_img) ; plt.imshow( pl_img ) #display

#File information display
print('size    : ', gray_img.size)
print('format  : ', gray_img.format) 
print('mode    : ', gray_img.mode) 
print('palette : ', gray_img.palette) 
print('info    : ', gray_img.info) 

#I think I have to do this
import matplotlib.cm as cm
plt.imshow(pl_img, cmap = cm.Greys_r)

Well, this time it's filtering

Built-in filter

Start with the built-in filter

`python`


from PIL import ImageFilter, ImageOps

`python`


#Blur
pl_img = np.array(gray_img.filter(ImageFilter.BLUR)); plt.imshow(pl_img, cmap = cm.Greys_r)

`python`


#Contour extraction
pl_img = np.array(gray_img.filter(ImageFilter.CONTOUR)); plt.imshow(pl_img, cmap = cm.Greys_r)

`python`


#Embossing
pl_img = np.array(filter_img = gray_img.filter(ImageFilter.EMBOSS)); plt.imshow(pl_img, cmap = cm.Greys_r)

`python`


#Minimum filter
pl_img = np.array(gray_img.filter(ImageFilter.MinFilter(5))); plt.imshow(pl_img, cmap = cm.Greys_r)

Own filter

I will write the theory later (Is it really written?)

`python`


#Detects vertical edges
flist =  [1,   1,  1,
          0,   0,  0,
         -1,  -1, -1]

flt = ImageFilter.Kernel((3, 3), flist, scale=1)
filter_img = gray_img.filter(flt)

pl_img = np.array(filter_img) ; plt.imshow(pl_img, cmap = cm.Greys_r)

`python`


#Detects horizontal edges
flist =  [1,   0, -1,
          1,   0, -1,
          1,   0, -1]
flt = ImageFilter.Kernel((3, 3), flist, scale=1)
filter_img = gray_img.filter(flt)

pl_img = np.array(filter_img) ; plt.imshow(pl_img, cmap = cm.Greys_r)

`python`


#4 Near proximity Laplacian
flist = [0,  1, 0,
         1, -4, 1,
         0,  1, 0]
flt = ImageFilter.Kernel((3, 3), flist, scale=1)
filter_img = gray_img.filter(flt)

pl_img = np.array(filter_img) ; plt.imshow(pl_img, cmap = cm.Greys_r)

`python`


#8 Near proximity Laplacian
flist = [1,  1, 1,
         1, -8, 1,
         1,  1, 1]
flt = ImageFilter.Kernel((3, 3), flist, scale=1)
filter_img = gray_img.filter(flt)

pl_img = np.array(filter_img) ; plt.imshow(pl_img, cmap = cm.Greys_r)

filter_img.save('work-image/filter05-lena.png')

`python`


#8 Example of using the nearby Laplacian
flist = [0.4,  0.4, 0.4,
         0.4, -2.2, 0.4,
         0.4,  0.4, 0.4]
flt = ImageFilter.Kernel((3, 3), flist, scale=1)
filter_img = gray_img.filter(flt)

pl_img = np.array(filter_img) ; plt.imshow(pl_img, cmap = cm.Greys_r)

↓ I gave a notebook to nbviewer (this is the main one) nbviewer.ipython.org/github/suto3/git-public/blob/master/python/notebook/Pillow-workflow03.ipynb

↓ Click here for the working environment Pillow environment construction --Virtual environment by virtualenv, interactive environment by iPython --Qiita

Try using Pillow on iPython (Part 1) --Qiita

Try using Pillow on iPython (Part 2) --Qiita