opencv_reference.py

'''
OpenCV Quick Reference
https://github.com/justmarkham/python-reference

By Stein Castillo (kevin@dataschool.io)
http://www.dataschool.io

Table of Contents:
    Imports
    Basic Image Operations
    Image Manipulation
'''

### IMPORTS ###

#Generic imports
import cv2
import cv2 as cv        # I use this import to make the calls easier.
import numpy as np 

#Import matplotlib to display images
from matplotlib import pyplot as plt 

#Extended mahotas functionality
import mahotas
import mahotas as mh    #typical import 

#Import if command line arguments are required
import argparse


### BASIC IMAGE OPERATIONS ###

#Load an image
image = cv2.imread(image)

#Load an image and convert to grayscale
image = cv2.imread(image, 0)
image = cv2.imread(image, cv2.IMREAD_GRAYSCALE) #both commands do the same thing

#Determine image properties
(h, w, c) = image.shape     #h: Height, w: Width, c: Channels
size = image.size           #Number of pixels
imgtype = image.dtype       #Image type
max = image.max()           #Maximum value
min = image.min()           #Minimum value

#Display an image with openCV
cv2.namedWindow(window_name, property)  #creates a window with specific properties:
                                        #cv2.WINDOW_NORMAL :User can resize the window (no constrain)
                                        #cv2.WINDOW_AUTOSIZE: Window size is automatically adjusted
                                        #cv2.WINDOW_OPENGL: Window created with OPENGL support


cv2.imshow(window_name, image)  #display image, automatic window properties


cv2.destroyWindow(window_name)  #Closes specified window
cv2.destroyAllWindows()         #Close all opencv windows


cv2.waitKey()                   #Display the windows and wait for user keypress
key = cv.waitKey(0) &0xFF       #Display the windows and store keypress 


#Display an image with MATPLOTLIB
img = cv2.imread(image, 0)
plt.imshow(img, cmap = 'gray', interpolation = 'bicubic')
plt.xticks([]), plt.yticks([])  # to hide tick values on X and Y axis
plt.show()


###IMAGE MANIPULATION###

#convert image to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

#Blur image

#Gaussian blur
#Very effective to remove gaussian noise in an image
#It is recommended to convert the image to grayscale before applying the blur filter.
blurred = cv2.GaussianBlur(image, (5, 5), 0)    #image: source image
                                                #(5, 5): Kernel size, must be positive and odd

#Average blur
#Used to smooth an image applying normalized box filter
#Average filter. takes the average of all the pixels under the kernel
blurred = cv2.blur(image, (5,5))        #image: Source image
                                        #(5,5): Kernel size, must be positive and odd

#Median blur
#Used to remove "salt and pepper" noise from an image
#Computes the median of all the pixels under the kernel
blurred = cv2.medianBlur(image, 5)      #image: Source image
                                        #5: Kernel size, most be positive and odd

#Applyng Gaussian blur filter with Mahotas
blurred = mahotas.gaussian_filter(image, 8) #imageö Source image
                                            #8: standard deviation for Gaussian kernel (in pixel units)



# Thresholding an image

""" Thresholding is the binarization of an image:
    any value over an index (C) is set to 255 (White)
    any value under the index (C) is set ot 0 (black)

    inverse binarizacion is also possible:
    any value over the index (C) is the to 0 (black)
    any value under the index (C) is set to 255 (white)"""

# Simple thresholding
# requires user intervetion to provide the index (C) value
(T, thresh) = cv2.threshold(src, C, 255, cv2.THRESH_BINARY) #src: source image
                                                            #C: index (as explained above)
                                                            #255: max value to assign if value greater than C
                                                            #Method:    cv2.THRESH_BINARY
                                                            #           cv2.THRESH_BINARY_INV
                                                            #Returns:   T: max value
                                                            #           thresh: Thresholded image
# Examples of simple thresholding
(T, thresh) = cv2.threshold(blurred, 155, 255, cv2.THRESH_BINARY)
(T, thresh) = cv2.threshold(blurred, 155, 255, cv2.THRESH_BINARY_INV)


# Resize an image
image = cv.resize(image, None, fx = 0.5, fy = 0.5, interpolation = cv.INTER_AREA)
"""
fx: set the resizing factor on x axis. In this example shrinks the image 50%
fy: set the resizing factor on y axis. In this example shrinks the image 50%
interpolation values:
cv.INTER_AREA : Used for image shrinking
cv.INTER_LINEAR: Used for zooming - Also used as default for all resizing purposes
cv.INTER_CUBIC: Also used for shrinking but is slower
""" 

# Change contrast and brightness of an image

result = cv.addWeighted(image, alpha, np.zeros(image.shape, image.dtype), beta, gamma)
"""
result = (image * alpha) + (np.zeros * beta) + gamma
gamma: changes the brightness of the image
alpha: changes the contrast of the image
beta: Does nothing since it is multipied to an array of zeros
alpha = 0.5, beta = 120 : X-ray effect
"""