Lab_4(c).py

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

def add_gaussian_noise(img, mean=0, stddev=100):
    gaussian_noise = np.random.normal(mean, stddev, img.shape)
    noisy_image = img + gaussian_noise
    noisy_image = np.clip(noisy_image, 0, 255)
    return noisy_image.astype(np.uint8)


def gaussian_high_pass_filter(image, cut_off_frequency):
    height, width = image.shape
    H = np.zeros(image.shape)
    D0 = cut_off_frequency

    frequency_domain_image = np.fft.fftshift(np.fft.fft2(image))

    for i in range(height):
        for j in range(width):
            D = np.sqrt((i - height / 2)**2 + (j-width/2) ** 2)
            H[i, j] = 1 - np.exp(-(D**2) / (2*D0)**2)

    filtered_image = frequency_domain_image * H
    filtered_image = np.abs(np.fft.ifft2(filtered_image))
    filtered_image = filtered_image / 255
    return filtered_image


def ideal_high_pass_filter(image, cut_off_frequency):
    height, width = image.shape
    H = np.zeros(image.shape)
    D0 = cut_off_frequency

    frequency_domain_image = np.fft.fftshift(np.fft.fft2(image))

    for i in range(height):
        for j in range(width):
            D = np.sqrt((i - height / 2)**2 + (j-width/2) ** 2)
            H[i, j] = D > D0

    filtered_image = frequency_domain_image * H
    filtered_image = np.abs(np.fft.ifft2(filtered_image))
    filtered_image = filtered_image / 255
    return filtered_image



path = 'Digital Image Processing/myph2.jpg'
img = cv2.imread(path,0)
img = cv2.resize(img,(512,512))

noisy_image = add_gaussian_noise(img,mean=0,stddev=40)

plt.figure(figsize=(8, 8))
number_of_filter = 16
cut_off_frequency = 5
dim = int(np.ceil(np.sqrt(number_of_filter)))

for i in range(number_of_filter):
    filtered_image1 = ideal_high_pass_filter(noisy_image, cut_off_frequency)
    filtered_image2 = gaussian_high_pass_filter(noisy_image, cut_off_frequency)

    plt.subplot(dim, dim, i+1)
    if i%2 ==0:
        plt.imshow(filtered_image1, cmap='gray')
        plt.title(f'Ideal high pass Filtered with D0 = {cut_off_frequency}')
    else:
        plt.imshow(filtered_image2, cmap='gray')
        plt.title(f'Gaussian high pass Filtered with D0 = {cut_off_frequency}')
    cut_off_frequency = cut_off_frequency + 5

plt.tight_layout()
plt.show()