denoise.py

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

# Read grayscale image
img = cv2.imread('image2.jpg', 0)
img = img.astype(np.float32)

# Padding for 3x3 kernel
padded = np.pad(img, 1, mode='reflect')

# Define output images
arithmetic = np.zeros_like(img)
geometric = np.zeros_like(img)
harmonic = np.zeros_like(img)
contra = np.zeros_like(img)

Q = 1.5  # Remove pepper noise

# Apply 3x3 neighborhood filters
for i in range(1, padded.shape[0] - 1):
    for j in range(1, padded.shape[1] - 1):
        window = padded[i-1:i+2, j-1:j+2]

        # Arithmetic Mean
        arithmetic[i-1, j-1] = np.mean(window)

        # Geometric Mean
        geometric[i-1, j-1] = np.exp(np.mean(np.log(window + 1e-8)))

        # Harmonic Mean
        harmonic[i-1, j-1] = 9 / np.sum(1.0 / (window + 1e-8))

        # Contra-harmonic Mean
        numerator = np.sum(window ** (Q + 1))
        denominator = np.sum(window ** Q + 1e-8)
        contra[i-1, j-1] = numerator / denominator

# Display results
titles = ["Original", "Arithmetic Mean", "Geometric Mean",
          "Harmonic Mean", "Contra-Harmonic Mean"]
images = [img, arithmetic, geometric, harmonic, contra]

plt.figure(figsize=(14, 10))
for i in range(5):
    plt.subplot(2, 3, i+1)
    plt.imshow(images[i], cmap="gray")
    plt.title(titles[i])
    plt.axis("off")

plt.tight_layout()
plt.show()