The image operation record of learning opencv

Calculating image may be a decimal value of binary arithmetic may be bitwise operations

An image adder

1 import cv2                                                                       
2 import numpy as np 
3 img1 = np.random.randint(0,256, size = [3,3] , dtype = np.uint8)
4 img2 = np.random.randint(0,256, size = [3,3] , dtype = np.uint8)
5 img3 = cv2.add(img1,img2) 
6 print("img1 =\n" , img1)
7 print("img2 =\n" , img2)
8 print("img1+img2 =\n" , img1 + img2)
9 print("cv2.add(img1, img2) =\n" , img3)

img1 = 
 [[ 106 40 179 ]
 [171 138 207]
 [131  34  55]]
img2 =
 [[125  73 224]
 [132 229 201]
 [151  37  40]]
img1 img2 = 
 [[ 231 113 147 ]
 [ 47 111 152]
 [ 26  71  95]]
cv2.add(img1, img2) =
 [[231 113 255]
 [255 255 255]
 [255  71  95]]

By adding the above code shows the addition operator and cv2.add () function

Addition operator

FIG two pixel values corresponding to the addition and less than 255, i.e., modulo 256, and

cv2.add () function

Will be the saturation value (maximum value) of the pixel values of two images, i.e., when greater than 255, 255. The other result is obtained and compared with the original values.

Operators two functions may be involved in computing two images, an image may be a value

Second, image weighting, and

I.e. weighted images and each image is a heavy weight into account, the formula is:

dst = saturate( src1 * a + src2 * b)

The results were the saturation value

1 import cv2                                                                       
2 dog =cv2.imread("/home/miao/dog.jpg")
3 cat =cv2.imread("/home/miao/cat.jpg")
4 result = cv2.addWeighted(dog , 0.6 , cat , 0.4 , 0) 
5 cv2.imshow("dog", dog)
6 cv2.imshow("cat" , cat)
7 cv2.imshow("result" , result)
8 cv2.waitKey()
9 cv2.destroyAllWindows()

Third, bitwise logical operations

Common bit operation function

Function name	The basic meaning of
cv2.bitwise_and()	Bitwise AND
cv2.bitwise_or()	Bitwise or
cv2.bitwise_xor()	Bitwise XOR
cv2.bitwise_nor	Bitwise

These functions may be added in the third parameter mask may be a mask calculation result obtained designated part

The calculation value is converted to a binary image is about further calculates,

Bitwise AND, any numerical operations are located at the press 0 0

According to this characteristic configuration mask image that only two values 0 and 255

The image and the mask image can be a bitwise mask image in an image portion designated

 1 import cv2                                                                                                                                       
 2 import numpy as np 
 3 a = np.random.randint(0 , 255, (5,5) , dtype = np.uint8)
 4 b = np.zeros((5,5) , dtype = np.uint8)
 5 b[0 : 3 , 0 : 3] = 255
 6 b[4,4] = 255
 7 c = cv2.bitwise_and(a,b)
 8 print("a = \n" , a) 
 9 print("b = \n" , b) 
10 print("c = \n" , c)

a =                                                                                                                                              
 [[215 164  27 216  90]                                                                                                                          
 [184  33  40 101 247]                                                                                                                           
 [ 77 150 189  26 251]
 [101  30  32  49  86]
 [ 64  86 175  28 221]]
b = 
 [[255 255 255   0   0]
 [255 255 255   0   0]
 [255 255 255   0   0]
 [  0   0   0   0   0]
 [  0   0   0   0 255]]
c = 
 [[215 164  27   0   0]
 [184  33  40   0   0]
 [ 77 150 189   0   0]
 [  0   0   0   0   0]
 [  0   0   0   0 221]]

Fourth, the bit plane decomposition

8-bit grayscale image, each pixel using 8-bit binary representation, the value can be represented by

value = a7*2^7 + a6*2^6 + a5*2^5 + a4*2^4 + a3*2^3 + a2*2^2 + a1*2^1 + a0*2^0

Weight ai weight of each of different effects on the image size are different

Thus it can be decomposed original 8-bit binary image, i.e., eight bit planes

a7 weights highest correlation with the picture of the largest, most similar to the image

0 smallest weight, the minimum configuration of image correlation, typically looks messy

Extracting a bit-plane binary image, the display image is directly up almost black image, because the pixel values of the low ground plane, an image near black display, so that after the extraction process value greater than the bit plane of bit 0 255, will be a binary image is displayed as a monochrome image

 1 import cv2                                                                                                                                       
 2 import numpy as np 
 3 dog = cv2.imread("/home/miao/dog.jpg" , 0) 
 4 cv2.imshow("dog" , dog)
 5 r,c = dog.shape
 6 x = np.zeros((r,c,8) , dtype = np.uint8)
 7 for i in range(8):
 8     x[:,:,i] = 2**i
 9 y = np.zeros((r,c,8) , dtype = np.uint8)
10 for i in range(8):
11     y[:,:,i] = cv2.bitwise_and(dog , x[:,:,i])
12     mask = y[:,:,i] > 0
13     y[mask] = 255
14     cv2.imshow(str(i) , y[:,:,i])
15 cv2.waitKey()
16 cv2.destroyAllWindows()

Original grayscale image

A0:

Fifth, the image encryption and decryption

By bitwise XOR encryption and decryption can be achieved

Exclusive OR operation is assumed by the rule:

xor(a,b) = c

Get

xor(c,b) = a

xor(c,a) = b

It can draw the following relationship

a: plaintext, the raw data

b: key

c: ciphertext

 1 import cv2                                                                                                                                       
 2 import numpy as np 
 3 dog = cv2.imread("/home/miao/dog.jpg" , 0) 
 4 r,c = dog.shape
 5 key = np.random.randint(0 , 256 , size = [r,c] , dtype = np.uint8)
 6 encryption = cv2.bitwise_xor( dog , key)
 7 decryption = cv2.bitwise_xor( encryption , key)
 8 cv2.imshow("dog" , dog)
 9 cv2.imshow("key" , key)
10 cv2.imshow("encryption" , encryption)
11 cv2.imshow("decryption" , decryption)
12 cv2.waitKey()
13 cv2.destroyAllWindows()