helpscriptpy.py

# -*- coding: utf-8 -*-
"""HelpScriptPy.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1_ZqIMIb-nZ87fAwkvPidCiefnPO1FmNN
"""

from os import listdir
from os.path import isfile, join
import numpy
import cv2
from scipy import signal
import glob
from PIL import Image
import os
import numpy as np
import tensorflow as tf
from os import listdir
from os.path import isfile, join
import matplotlib.pyplot as plt
from skimage import exposure

#Function to load data as particular type in array
def load_data_astype(mypath, type):
    onlyfiles = [ f for f in listdir(mypath) if isfile(join(mypath,f)) ]
    images = numpy.empty(len(onlyfiles), dtype=object)
    for n in range(0, len(onlyfiles)):
      images[n] = cv2.imread( join(mypath,onlyfiles[n]), type )
    return images

#Function to load data in array
def load_data(mypath):
    onlyfiles = [ f for f in listdir(mypath) if isfile(join(mypath,f)) ]
    images = numpy.empty(len(onlyfiles), dtype=object)
    for n in range(0, len(onlyfiles)):
      images[n] = cv2.imread( join(mypath,onlyfiles[n]))
    return images

#Function to display array of images
def displayImageArray(image, title):
    for n in range(0, len(image)):
      plt.imshow(cv2.cvtColor(image[n],cv2.COLOR_BGR2RGB))
      plt.title(title)
      plt.show()  

#Function to display single image
def displayImage(image, title):
    plt.imshow(cv2.cvtColor(image,cv2.COLOR_BGR2RGB))
    plt.title(title)
    plt.show()

#Function to display single image as Gray Scale
def displayGrayImage(image, title):
    plt.imshow(cv2.cvtColor(image,cv2.COLOR_BGR2GRAY))
    plt.title(title)
    plt.show()

#Function to exclude particular channel from color image and display image with remaining channels
def rgbExclusion(image, type):
    if 'red' in type: 
      red_image = image.copy()
      red_image[:,:,2] = 0 #empty red channel
      displayImage(red_image,"Excluding Red Channel")
    if 'green' in type: 
      green_image = image.copy()
      green_image[:,:,1] = 0 #empty green channel
      displayImage(green_image,"Excluding Green Channel")
    if 'blue' in type: 
      blue_image = image.copy()
      blue_image[:,:,0] = 0 #empty blue channel
      displayImage(blue_image,"Excluding Blue Channel")

#Function to plot histogram (simple or equilized depending upon which parameter is passed) and compute cdf
def plotHistogram(image, type):
    hist,bins = np.histogram(image.flatten(),256,[0,256])
    
    cdf = hist.cumsum()
    cdf_normalized = cdf * hist.max()/ cdf.max()
    
    plt.plot(cdf_normalized, color = 'b')
    if (type==0):
      plt.title("Histogram of Image")
    if (type==1):
      plt.title("Histogram Equilization of Image")
    plt.hist(image.flatten(),256,[0,256], color = 'r')
    plt.xlim([0,256])
    plt.legend(('cdf','histogram'), loc = 'upper left')
    plt.show()
    
#Function to apply histogram equilization on image
def histogramEquilization(image):
    hist,bins = np.histogram(image.flatten(),256,[0,256])
    cdf = 0
    cdf = hist.cumsum()
    cdf_normalized = cdf * hist.max()/ cdf.max()
    cdf_m = np.ma.masked_equal(cdf,0)
    cdf_m = (cdf_m - cdf_m.min())*255/(cdf_m.max()-cdf_m.min())
    cdf = np.ma.filled(cdf_m,0).astype('uint8')
    img2 = cdf[image]
    equ = cv2.equalizeHist(image)
    return equ

#Function to implement Convolution from scratch using image and kernel
def Convolve2dfromScratch(image, kernel, type):
    kernel = np.flipud(np.fliplr(kernel))    # Flip the kernel
    output = np.zeros_like(image)            # convolution output
    # Add zero padding to the input image
    image_padded = np.zeros((image.shape[0] + 2, image.shape[1] + 2))   
    image_padded[1:-1, 1:-1] = image
    
    # Loop over every pixel of the image and implement convolution operation (element wise multiplication and summation). 
    # You can use two loops. The result is stored in the variable output.
    
    for x in range(image.shape[0]):     # Loop over every pixel of the image
        for y in range(image.shape[1]):
            # element-wise multiplication and summation 
            output[x,y]=(kernel*image_padded[x:x+3,y:y+3]).sum()
    return output

#Function to implement convolution using library function signal.convolve2d
def Convolve2dUsingLibrary(image, kernel, type):
      image = signal.convolve2d(image,kernel, 'same')
      return image

#Function to show comparison of both convolutions (manual and library)
def displayComaprisonConvolution(image):
      fig, (ax1, ax2) = plt.subplots(1, 2)
      fig.suptitle('Blur Image')
      ax1.imshow(image[0], cmap=plt.cm.gray)
      ax1.set_title('ManualConv')
      ax2.imshow(image[3], cmap=plt.cm.gray)
      ax2.set_title('ConvUsingLib')

      fig, (ax1, ax2) = plt.subplots(1, 2)
      fig.suptitle('Blurrer Image')
      ax1.imshow(image[1], cmap=plt.cm.gray)
      ax1.set_title('ManualConv')
      ax2.imshow(image[4], cmap=plt.cm.gray)
      ax2.set_title('ConvUsingLib')

      fig, (ax1, ax2) = plt.subplots(1, 2)
      fig.suptitle('Sharp Image')
      ax1.imshow(image[2], cmap=plt.cm.gray)
      ax1.set_title('ManualConv')
      ax2.imshow(image[5], cmap=plt.cm.gray)
      ax2.set_title('ConvUsingLib')

#Function to apply Box Filter given image and kernel size
def BoxFilter(imagebox, type):
    if (type == 3):
      kernel = np.ones((type,type),np.float32)/9 #blur
    if (type == 9):
      kernel = np.ones((type,type),np.float32)/25 #blur
    image = signal.convolve2d(imagebox,kernel, 'same')
    return image

#Function to apply Guassian Filter given image and sigma
def GuassianFilter(image, sigma):
    image = cv2.GaussianBlur(image, (5,5),sigmaX=sigma)
    return image

#Function to apply Median Filter given image
def MedianFilter(image):
    image = signal.medfilt(image, (5,5))
    return image

#Function to apply noise i.e. guassian and salt and pepper to image
def Noisy(noise_typ, image):
   if noise_typ == "gauss":
      mean = 0.0   # some constant
      std = 5.0    # some constant (standard deviation)
      noisy_img = image + np.random.normal(mean, std, image.shape)
      noisy_img_clipped = np.clip(noisy_img, 0, 255) 
      return noisy_img_clipped
   elif noise_typ == "s&p":
      row,col = image.shape
      s_vs_p = 0.5
      amount = 0.004
      out = np.copy(image)
      # Salt mode
      num_salt = np.ceil(amount * image.size * s_vs_p)
      coords = [np.random.randint(0, i - 1, int(num_salt))
              for i in image.shape]
      out[coords] = 1

      # Pepper mode
      num_pepper = np.ceil(amount* image.size * (1. - s_vs_p))
      coords = [np.random.randint(0, i - 1, int(num_pepper))
              for i in image.shape]
      out[coords] = 0
      return out

#Function to draw mesh plot first order derivative in X direction with sigma as parameter
def first_dev_guassian_x(x, y, sigma):
    temp = (x ** 2 + y ** 2) / (2 * sigma ** 2)
    return -x * np.exp(-temp)/ (2* np.pi * sigma ** 4) 

#Function to draw mesh plot first order derivative in Y direction with sigma as parameter
def first_dev_guassian_y(x, y, sigma):
    temp = (x ** 2 + y ** 2) / (2 * sigma ** 2)
    return -y * np.exp(-temp)/ (2* np.pi * sigma ** 4)

#Function to draw mesh plot second order derivative in X direction with sigma as parameter
def second_dev_guassian_x(x, y, sigma):
    temp = (x ** 2 + y ** 2) / (2 * sigma ** 2)
    temp1 = np.exp(-temp) /(2* np.pi * sigma ** 4) 
    return (-1 + x**2 / sigma**2)* temp1

#Function to draw mesh plot second order derivative in Y direction with sigma as parameter
def second_dev_guassian_y(x, y, sigma):
    temp = (x ** 2 + y ** 2) / (2 * sigma ** 2)
    temp1 = np.exp(-temp) /(2* np.pi * sigma ** 4) 
    return (-1 + y**2 / sigma**2)* temp1

#Function to compute 3d mesh plot for guassian filter
def guassian(x, y, sigma):
    temp = (x ** 2 + y ** 2) / (2 * sigma ** 2)
    return -1 * np.exp(-temp)/ (2* np.pi * sigma ** 2) 

#Function to draw mesh plot second order derivative with sigma as parameter
def second_dev_guassian(x, y, sigma):
    temp = (x ** 2 + y ** 2) / (2 * sigma ** 2)
    temp1 = np.exp(-temp) /(2* np.pi * sigma ** 4) 
    return (x * y / 2 * np.pi *sigma**6)* temp1

#Function to apply and plot Sobel Operator given image and kernel size
def SobelOperator(image, index, size):
    #Sobel X
    img_SobelX =  cv2.Sobel(image,cv2.CV_64F,1,0,ksize=size)

    #Sobel Y 
    img_SobelY = cv2.Sobel(image,cv2.CV_64F,0,1,ksize=size)

    #Gradient Magnitude
    gradient_magnitude = np.sqrt((np.power(img_SobelX,2)+np.power(img_SobelY,2)))

    fig, (ax0, ax1, ax2, ax3) = plt.subplots(1, 4, figsize=(15,15))
    if size==3:
      ax0.imshow(image, cmap=plt.cm.gray)
      ax0.set_title('Original Image')
      ax1.imshow(img_SobelX, cmap=plt.cm.gray)
      ax1.set_title('Sobel X kernel size 3 ')
      ax2.imshow(img_SobelY, cmap=plt.cm.gray)
      ax2.set_title('Sobel Y kernel size 3')
      ax3.imshow(gradient_magnitude, cmap=plt.cm.gray)
      ax3.set_title('Gradient Magnitude kernel size 3 ')
    if size==9:
      ax0.imshow(image, cmap=plt.cm.gray)
      ax0.set_title('Original Image')
      ax1.imshow(img_SobelX, cmap=plt.cm.gray)
      ax1.set_title('Sobel X kernel size 9 ')
      ax2.imshow(img_SobelY, cmap=plt.cm.gray)
      ax2.set_title('Sobel Y kernel size 9')
      ax3.imshow(gradient_magnitude, cmap=plt.cm.gray)
      ax3.set_title('Gradient Magnitude kernel size 9 ')
    
#Function to apply and plot Laplacian Operator given image
def Laplacian(image, index):
    low_sigma = cv2.GaussianBlur(image,(3,3),1)
    high_sigma = cv2.GaussianBlur(image,(5,5),8)
 
    # Calculate the DoG by subtracting
    laplacian = low_sigma - high_sigma

    fig, (ax0, ax1, ax2, ax3) = plt.subplots(1, 4, figsize=(15,15))
    ax0.imshow(image, cmap=plt.cm.gray)
    ax0.set_title('Original Image')
    ax1.imshow(high_sigma, cmap=plt.cm.gray)
    ax1.set_title('Guassian with sigma 8')
    ax2.imshow(low_sigma, cmap=plt.cm.gray)
    ax2.set_title('Guassian with sigma 1')
    ax3.imshow(laplacian, cmap=plt.cm.gray)
    ax3.set_title('Laplacian Image')
    return laplacian

#Function to apply and plot Canny Operator given image
def CannyEdgeDetector(image):
    edges = cv2.Canny(image, 100, 200)
    plt.rcParams['figure.figsize'] = (20.0, 16.0) # set default size of plots
    fig = plt.figure()
    fig.add_subplot(1, 3, 1)
    plt.imshow(image,cmap=plt.cm.gray)
    plt.title("GrayScale")
    fig.add_subplot(1, 3, 2).imshow(edges,cmap=plt.cm.gray)
    plt.title("Canny edge detector")