def kmeanDraw(self):
cp_imgarr = copy.deepcopy(self.px)
img = Image.new('RGB', (self.img_width, self.img_height), "white")
p = img.load()
Distance = cdist(self.px, self.centroids)
cluster_id = np.argmin(Distance, axis=1)
for i in range(len(cluster_id)):
RGB_value = self.centroids[cluster_id[i]] * 255
cp_imgarr[i] = RGB_value
i = 0
for x in range(self.img_width):
for y in range(self.img_height):
p[x, y] = tuple(cp_imgarr[i].astype(int))
i += 1
img.show()
img.save('Kmean' + str(self.K) + '_poblem3.jpg')
def loadImage(nombreImage='leon.jpeg', mostrarImagen=True):
img = Image.open(nombreImage)
if (mostrarImagen == True):
img.show()
return img
def taskC():
image = Image.open('f1.jpg')
print(image.size)
image.thumbnail((100, 100))
print(image.size)
image.show()
def taskA():
image = Image.open("f1.jpg")
print(image.format)
print(image.mode)
print(image.size)
image.show()
'''
############################### Method 3: Keras: ###############################
'''
print("##### Keras #####")
################################ Load image ###############################
from keras.preprocessing.image import load_img
# load the image in PIL format
image = load_img("3096_color.jpeg")
# report details about the image
print(type(image)) # <class 'PIL.JpegImagePlugin.JpegImageFile'>
print(image.format) # JPEG
print(image.mode) # RGB
print(image.size) # (481, 321)
image.show()
################################ Convert image PIL into np.array ###############################
from keras.preprocessing.image import img_to_array
image_array = img_to_array(image)
print(type(image_array)) # <class 'numpy.ndarray'>
print(image_array.dtype) # float32
print(image_array.shape) # (321, 481, 3)
################################ Reverse np.array into PIL ###############################
from keras.preprocessing.image import array_to_img
image_PIL = array_to_img(image_array)
print(type(image_PIL)) # <class 'PIL.Image.Image'>
################################ Save image ###############################
from keras.preprocessing.image import save_img