In [1]:

    

import cv2
import numpy as np
import matplotlib.pyplot as plt

%matplotlib inline


    

In [2]:

    

!curl "https://reviewonline.co.za/wp-content/uploads/sites/68/2018/12/giraffes1.jpg" -o /tmp/img.jpg


    

    




  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100  167k  100  167k    0     0   250k      0 --:--:-- --:--:-- --:--:--  250k

In [3]:

    

img = cv2.imread("/tmp/img.jpg")


    

In [4]:

    

img.shape


    

    
Out[4]:





(514, 983, 3)

In [5]:

    

img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)


    

In [6]:

    

plt.imshow(img)


    

    
Out[6]:





<matplotlib.image.AxesImage at 0x120470510>






    

In [7]:

    

img.shape


    

    
Out[7]:





(514, 983, 3)

In [8]:

    

fig, axes = plt.subplots(1, 3, figsize = (15,3))


img1 = img.copy()
img1[:,:,1]=0
img1[:,:,2]=0
axes[0].imshow(img1)

img1 = img.copy()
img1[:,:,0]=0
img1[:,:,2]=0
axes[1].imshow(img1)

img1 = img.copy()
img1[:,:,1]=0
img1[:,:,0]=0
axes[2].imshow(img1)


    

    
Out[8]:





<matplotlib.image.AxesImage at 0x120b5e990>






    

In [9]:

    

plt.hist(img[:,:,0].flatten(), bins = 50, histtype = "step", color = "red", density=True)
plt.hist(img[:,:,1].flatten(), bins = 50, histtype = "step", color = "green", density=True)
plt.hist(img[:,:,2].flatten(), bins = 50, histtype = "step", color = "blue", density=True)
plt.xlabel("Pixel")
plt.ylabel("Density");


    

In [10]:

    

img5 = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
plt.imshow(img5, cmap="gray")


    

    
Out[10]:





<matplotlib.image.AxesImage at 0x120d90410>






    

In [11]:

    

plt.hist(img5.flatten(), bins = 50, histtype = "step", color = "gray", density=True);


    

In [12]:

    

height, width = img.shape[:2]
height, width


    

    
Out[12]:





(514, 983)

In [13]:

    

img2 = cv2.resize(img,(width * 2, height * 2), interpolation = cv2.INTER_CUBIC)
print(img2.shape)

plt.figure(figsize = (10, 8))
plt.imshow(img2)


    

    




(1028, 1966, 3)






    
Out[13]:





<matplotlib.image.AxesImage at 0x1233d5e50>






    

In [14]:

    

img2 = cv2.resize(img,(int(width/2), int(height/2)), interpolation = cv2.INTER_CUBIC)
print(img2.shape)

plt.figure(figsize = (10, 8))
plt.imshow(img2)


    

    




(257, 491, 3)






    
Out[14]:





<matplotlib.image.AxesImage at 0x123331110>






    

In [15]:

    

img3 = cv2.GaussianBlur(img, ksize=(31, 31), sigmaX = 10.0, sigmaY = 10.0)
plt.imshow(img3)
plt.title("Gaussian Blur");


    

In [16]:

    

fig, axes = plt.subplots(1, 4, figsize = (15,3))

axes[0].imshow(img)
axes[0].set_title("Origin")
axes[1].imshow(cv2.blur(img, (10, 10)))
axes[1].set_title("Blur - 10x10")
axes[2].imshow(cv2.blur(img, (20, 20)))
axes[2].set_title("Blur - 20x20")
axes[3].imshow(cv2.blur(img, (30, 30)))
axes[3].set_title("Blur - 30x30");


    

In [17]:

    

plt.imshow(img[70:270,50:250,:])


    

    
Out[17]:





<matplotlib.image.AxesImage at 0x123f65550>






    

In [18]:

    

img3 = img.copy()
img3[70:270,50:250,:] = 255
plt.imshow(img3)


    

    
Out[18]:





<matplotlib.image.AxesImage at 0x123391790>






    

In [19]:

    

img3 = img.copy()
img3[70:270,50:250,:] = 0
plt.imshow(img3)


    

    
Out[19]:





<matplotlib.image.AxesImage at 0x12356cf90>






    

In [20]:

    

img3 = img.copy()
img3[70:270,50:250,:] = cv2.blur(img3[70:270,50:250,:], ksize=(20,20))
plt.imshow(img3)
plt.title("Masked with blurred rectangle")


    

    
Out[20]:





Text(0.5, 1.0, 'Masked with blurred rectangle')






    

In [21]:

    

img3 = img.copy()
img3[70:270,50:250,:] = cv2.blur(img3[70:270,50:250,:], ksize=(20,20))
cv2.rectangle(img3, (50, 70), (250, 270), (255, 255, 255), 2)
plt.imshow(img3)
plt.title("Masked with transparent white rectangle")


    

    
Out[21]:





Text(0.5, 1.0, 'Masked with transparent white rectangle')






    

In [22]:

    

img3 = img.copy()
cv2.circle(img3, (100, 180), 100, (255, 255, 255), 3)
plt.imshow(img3)
plt.title("Masked with transparent white circle")


    

    
Out[22]:





Text(0.5, 1.0, 'Masked with transparent white circle')






    

In [23]:

    

img1 = img.copy()
cv2.circle(img1, (100, 180), radius = 100, color = (255, 255, 255), thickness = -1)
plt.imshow(img1)
plt.title("Masked with white circle")


    

    
Out[23]:





Text(0.5, 1.0, 'Masked with white circle')






    

In [24]:

    

img1 = img.copy()
cv2.rectangle(img1, pt1 = (50, 70), pt2 = (250, 270), color = (255, 255, 255), thickness = -1)
plt.imshow(img1)
plt.title("Masked with white rectangle")


    

    
Out[24]:





Text(0.5, 1.0, 'Masked with white rectangle')






    

In [25]:

    

mask = cv2.circle(np.zeros_like(img), (100, 180), radius = 100, color = (255, 255, 255), thickness = -1)
img3 = cv2.bitwise_and(img, mask)
plt.imshow(img3)
plt.title("Bitwise masked")


    

    
Out[25]:





Text(0.5, 1.0, 'Bitwise masked')






    

In [26]:

    

M = np.float32([[1,0,100],[0,1,50]])
img4 = cv2.warpAffine(img, M, (width,height))
plt.imshow(img4)
plt.title("Shifted")


    

    
Out[26]:





Text(0.5, 1.0, 'Shifted')






    

In [27]:

    

fig, axes = plt.subplots(3, 2, figsize = (10,12))

axes[0][0].imshow(img)
axes[0][1].imshow(cv2.flip(img, flipCode = 1))
axes[0][0].set_title("Original")
axes[0][1].set_title("Flipped Horizontally")

axes[1][0].imshow(img)
axes[1][1].imshow(cv2.flip(img, flipCode = 0))
axes[1][0].set_title("Original")
axes[1][1].set_title("Flipped Vertically")

axes[2][0].imshow(img)
axes[2][1].imshow(cv2.flip(img, flipCode = -1))
axes[2][0].set_title("Original")
axes[2][1].set_title("Flipped Horizontally and Vertically")


    

    
Out[27]:





Text(0.5, 1.0, 'Flipped Horizontally and Vertically')






    

In [28]:

    

w, h = img.shape[:2]
rotation_matrix = cv2.getRotationMatrix2D(center = (h//2, w//2), angle = 10, scale = 1.0)
fig, axes = plt.subplots(1, 2, figsize = (15,3))
axes[0].imshow(img)
axes[0].set_title("Original")
axes[1].imshow(cv2.warpAffine(img, rotation_matrix, (h, w)))
axes[1].set_title("Rotate 10 degree")


    

    
Out[28]:





Text(0.5, 1.0, 'Rotate 10 degree')






    

In [29]:

    

fig, axes = plt.subplots(1, 4, figsize = (15,3))

img1 = img.copy()
axes[0].imshow(img)
axes[0].set_title("Original")
axes[1].imshow(cv2.rotate(img1, cv2.ROTATE_90_CLOCKWISE))
axes[1].set_title("Rotated +90")
axes[2].imshow(cv2.rotate(img1, cv2.ROTATE_90_COUNTERCLOCKWISE))
axes[2].set_title("Rotated -90")
axes[3].imshow(cv2.rotate(img1, cv2.ROTATE_180))
axes[3].set_title("Rotated +180")


    

    
Out[29]:





Text(0.5, 1.0, 'Rotated +180')






    

In [30]:

    

fig, axes = plt.subplots(1, 2, figsize = (15,3))

hsvImg = cv2.cvtColor(img, cv2.COLOR_RGB2HSV)
hsvImg[...,2] = hsvImg[...,2] * 0.7
axes[0].imshow(img)
axes[0].set_title("Original")
axes[1].imshow(cv2.cvtColor(hsvImg, cv2.COLOR_HSV2RGB))
axes[1].set_title("Decreased brightness")


    

    
Out[30]:





Text(0.5, 1.0, 'Decreased brightness')






    

In [31]:

    

fig, axes = plt.subplots(1, 2, figsize = (15,3))

hsvImg = cv2.cvtColor(img, cv2.COLOR_RGB2HSV)

value = 50

vValue = hsvImg[...,2]
hsvImg[...,2] = np.where((255 - vValue) < value, 255, vValue + value)

axes[0].imshow(img)
axes[0].set_title("Original")
axes[1].imshow(cv2.cvtColor(hsvImg, cv2.COLOR_HSV2RGB))
axes[1].set_title("Increased brightness")


    

    
Out[31]:





Text(0.5, 1.0, 'Increased brightness')






    

In [ ]: