Getting Started with Python and openCV

Standar Test Images

Images used in the exampes



In [24]:

    
#print the version of openCV
%matplotlib inline 
import cv2
import numpy as np
from matplotlib import pyplot as plt

print(cv2.__version__)



In [25]:

    
img = cv2.imread("./images/fubar.jpg")



In [26]:

    
plt.imshow(img)
plt.show()

Loading with different color modes

bgr vs rgb

Different programs need the color in different order opencv uses: BGR

cv.imread() and the colors need to be: matplotlib expects the colors in this sequence: RGB

This can be converted for display by using a color converter function: cv2.cvtColor(image, cv2.COLOR_BGR2RGB)



In [27]:

    
imgColorFix = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
plt.imshow(imgColorFix)
plt.show()



In [28]:

    
#let's try to clean up the display
plt.imshow(imgColorFix,interpolation="nearest")
plt.show()



In [29]:

    
#We can resize the image
imgResized = cv2.resize(imgColorFix, None, fx=1/10, fy=1/10)
plt.imshow(imgResized)
plt.show()



In [30]:

    
#Let's try to clean up the display
plt.imshow(imgResized, interpolation="nearest")
plt.show()



In [47]:

    
imgResized.shape









    Out[47]:





(136, 205, 3)



In [ ]:

    
imgResized



In [31]:

    
#Let's try printing the color constants for imread
print(cv2.IMREAD_GRAYSCALE)



In [32]:

    
#It's possible to read in images as different kinds with different color options
imgGray = cv2.imread("./images/fubar.jpg", cv2.IMREAD_GRAYSCALE)



In [ ]:



In [33]:

    
#The trouble here is that we have grayscale on one channel but matplotlib doesn't know what channel to assign it too
plt.imshow(imgGray)
plt.show()



In [34]:

    
#We can force matplot lib to display grayscale
plt.imshow(imgGray, cmap=plt.cm.gray)
plt.show()



In [45]:

    
imgGray.shape









    Out[45]:





(1362, 2048)

show red channel as red

So the issue is that matplotlib wasn't to display a complete image. We need to know:

What is a complete image?
How is it represented?
How can we get a read channel to show as red?

What is an image?

64x64 color test image

In opencv it's an array of colors, and pixels.



In [35]:

    
#load the test image and show it
img = cv2.imread("./images/colortest.png")
imgColorFix = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
plt.imshow(imgColorFix)
plt.show()



In [36]:

    
#What is the shape?
img.shape









    Out[36]:





(104, 158, 3)



In [37]:

    
#color fixed for matplotlib shape
imgColorFix.shape









    Out[37]:





(104, 158, 3)



In [ ]:



In [38]:

    
#There the same shape but the color order in the 3rd position is different
#The image is not centered around the color data we can fix that 
img_cropped = imgColorFix[28:92, 72:136]
plt.imshow(img_cropped, interpolation="nearest")
plt.show()



In [39]:

    
#by changing the shape dimensions the image shape changed
img_cropped.shape









    Out[39]:





(64, 64, 3)



In [40]:

    
#Seperating the color channels
images = []
for ii in [0,1,2]:
    color = img_cropped.copy()
    #empty the color channels
    if ii != 0: color[:,:,0] = 0
    if ii != 1: color[:,:,1] = 0
    if ii != 2: color[:,:,2] = 0
    images.append(color)   
    
plt.imshow(np.vstack(images), interpolation="nearest")
plt.show()



In [41]:

    
# Now we can just show the red channel
plt.imshow(images[0])
plt.show()



In [42]:

    
#There are several other manipulations we can too

image_blur = cv2.GaussianBlur(img_cropped, (7,7),0)
plt.imshow(image_blur);
plt.show()

More Resources

https://alexlouden.com/posts/2016-computer-vision-on-strawberries.html