4/16/25, 10:08 PM Untitled12.

ipynb - Colab

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

# Load grayscale image

image = cv2.imread('br.jpg', cv2.IMREAD_GRAYSCALE) # Replace with your image path

# 1. Strong Gaussian Noise

def add_gaussian_noise(img, mean=0, sigma=50):
gaussian = np.random.normal(mean, sigma, img.shape).astype(np.float32)
noisy_img = np.clip(img.astype(np.float32) + gaussian, 0, 255).astype(np.uint8)
return noisy_img

# 2. Strong Salt & Pepper Noise

def add_salt_pepper_noise(img, amount=0.1, salt_vs_pepper=0.5):
noisy_img = img.copy()
num_salt = int(np.ceil(amount * img.size * salt_vs_pepper))
num_pepper = int(np.ceil(amount * img.size * (1.0 - salt_vs_pepper)))

# Salt noise
coords = [np.random.randint(0, i - 1, num_salt) for i in img.shape]
noisy_img[tuple(coords)] = 255

# Pepper noise
coords = [np.random.randint(0, i - 1, num_pepper) for i in img.shape]
noisy_img[tuple(coords)] = 0

return noisy_img

# 3. Uneven Illumination (Vignette Effect)

def apply_vignette(img, strength=100):
rows, cols = img.shape
kernel_x = cv2.getGaussianKernel(cols, strength)
kernel_y = cv2.getGaussianKernel(rows, strength)
kernel = kernel_y @ kernel_x.T
mask = kernel / np.max(kernel) # Normalize to [0,1]
vignette_img = img * mask
vignette_img = np.clip(vignette_img, 0, 255).astype(np.uint8)
return vignette_img

# Apply the transformations

gaussian_noise = add_gaussian_noise(image)
sp_noise = add_salt_pepper_noise(image)
vignette = apply_vignette(image)

# Show results
plt.figure(figsize=(12, 6))
plt.subplot(1, 4, 1), plt.imshow(image, cmap='gray'), plt.title('Original'), plt.axis('off')
plt.subplot(1, 4, 2), plt.imshow(gaussian_noise, cmap='gray'), plt.title('Gaussian Noise'), plt.axis('off')
plt.subplot(1, 4, 3), plt.imshow(sp_noise, cmap='gray'), plt.title('Salt & Pepper'), plt.axis('off')
plt.subplot(1, 4, 4), plt.imshow(vignette, cmap='gray'), plt.title('Uneven Illumination'), plt.axis('off')
plt.tight_layout()
plt.show()

https://colab.research.google.com/drive/1c-YnbIsIZJ8J0JYjSYZDOKnjdcUwwpVs#scrollTo=KOkv9qjbdGBc&printMode=true 1/9
4/16/25, 10:08 PM Untitled12.ipynb - Colab

 

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

# Load grayscale image

image = cv2.imread('br.jpg', cv2.IMREAD_GRAYSCALE) # Replace with your actual image path

# Functions for noise and vignette

def add_gaussian_noise(img, mean=0, sigma=50):
noise = np.random.normal(mean, sigma, img.shape).astype(np.float32)
return np.clip(img.astype(np.float32) + noise, 0, 255).astype(np.uint8)

def add_salt_pepper_noise(img, amount=0.1, salt_vs_pepper=0.5):

noisy_img = img.copy()
num_salt = int(amount * img.size * salt_vs_pepper)
num_pepper = int(amount * img.size * (1 - salt_vs_pepper))
coords = [np.random.randint(0, i - 1, num_salt) for i in img.shape]
noisy_img[tuple(coords)] = 255
coords = [np.random.randint(0, i - 1, num_pepper) for i in img.shape]
noisy_img[tuple(coords)] = 0
return noisy_img

def apply_vignette(img, strength=100):

rows, cols = img.shape
kernel_x = cv2.getGaussianKernel(cols, strength)
kernel_y = cv2.getGaussianKernel(rows, strength)
mask = kernel_y @ kernel_x.T
mask = mask / np.max(mask)
vignette = img * mask
return np.clip(vignette, 0, 255).astype(np.uint8)

# Apply modifications
gaussian_img = add_gaussian_noise(image)
sp_img = add_salt_pepper_noise(image)
vignette_img = apply_vignette(image)

# Function to plot histogram

def plot_histogram(img, title, bins=64):
plt.hist(img.ravel(), bins=bins, range=(0, 256), color='gray')
plt.title(f'{title}')
plt.xlabel('Pixel Intensity')
plt.ylabel('Frequency')

# Plot all histograms

plt.figure(figsize=(12, 6))
plt.subplot(2, 2, 1)
plot_histogram(image, 'Original Image Histogram')

plt.subplot(2, 2, 2)
plot_histogram(gaussian_img, 'Gaussian Noise Histogram')

plt subplot(2 2 3)
https://colab.research.google.com/drive/1c-YnbIsIZJ8J0JYjSYZDOKnjdcUwwpVs#scrollTo=KOkv9qjbdGBc&printMode=true 2/9
4/16/25, 10:08 PM Untitled12.ipynb - Colab
plt.subplot(2, 2, 3)
plot_histogram(sp_img, 'Salt & Pepper Histogram')

plt.subplot(2, 2, 4)
plot_histogram(vignette_img, 'Uneven Illumination Histogram')

plt.tight_layout()
plt.show()

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

# Load grayscale and color version of the image

image_path = 'br.jpg' # Replace with actual image path
image_color = cv2.imread(image_path)
image_gray = cv2.cvtColor(image_color, cv2.COLOR_BGR2GRAY)

# 1. Basic Global Thresholding (Fixed threshold at 127)

_, basic_thresh = cv2.threshold(image_gray, 127, 255, cv2.THRESH_BINARY)

# 2. Otsu's Thresholding (automatic)

_, otsu_thresh = cv2.threshold(image_gray, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)

# 3. Multispectral Thresholding (each RGB channel separately)

b_channel, g_channel, r_channel = cv2.split(image_color)

# Apply threshold to each channel

_, b_thresh = cv2.threshold(b_channel, 127, 255, cv2.THRESH_BINARY)
_, g_thresh = cv2.threshold(g_channel, 127, 255, cv2.THRESH_BINARY)
_, r_thresh = cv2.threshold(r_channel, 127, 255, cv2.THRESH_BINARY)

# Combine using logical OR

multi_thresh = cv2.bitwise_or(cv2.bitwise_or(r_thresh, g_thresh), b_thresh)

# Display all outputs

plt.figure(figsize=(12, 8))

plt.subplot(2, 2, 1)
plt.imshow(image_gray, cmap='gray')
plt.title('Original Grayscale')
l i (' ff')
https://colab.research.google.com/drive/1c-YnbIsIZJ8J0JYjSYZDOKnjdcUwwpVs#scrollTo=KOkv9qjbdGBc&printMode=true 3/9
4/16/25, 10:08 PM Untitled12.ipynb - Colab
plt.axis('off')

plt.subplot(2, 2, 2)
plt.imshow(basic_thresh, cmap='gray')
plt.title('Basic Global Thresholding')
plt.axis('off')

plt.subplot(2, 2, 3)
plt.imshow(otsu_thresh, cmap='gray')
plt.title("Otsu's Thresholding")
plt.axis('off')

plt.subplot(2, 2, 4)
plt.imshow(multi_thresh, cmap='gray')
plt.title('Multispectral Thresholding (RGB)')
plt.axis('off')

plt.tight_layout()
plt.show()

import cv2
import numpy as np
import matplotlib.pyplot as plt
from skimage.segmentation import felzenszwalb, flood
https://colab.research.google.com/drive/1c-YnbIsIZJ8J0JYjSYZDOKnjdcUwwpVs#scrollTo=KOkv9qjbdGBc&printMode=true 4/9
4/16/25, 10:08 PM Untitled12.ipynb - Colab
from skimage.segmentation import felzenszwalb, flood
from skimage import img_as_float

# ----------------------------
# 1. Load & Resize Image
# ----------------------------
image_path = 'br.jpg' # 🔁 Replace with your file
image_color = cv2.imread(image_path)

if image_color is None:
raise FileNotFoundError("⚠️ Image not found. Check your path or upload again.")

resized_image = cv2.resize(image_color, (256, 256))

resized_float = img_as_float(resized_image)

# ----------------------------
# 2. Felzenszwalb Segmentation
# ----------------------------
felz_segments = felzenszwalb(resized_float, scale=100, sigma=0.5, min_size=50)

# ----------------------------
# 3. Region Growing (auto seed)
# ----------------------------
gray_resized = cv2.cvtColor(resized_image, cv2.COLOR_BGR2GRAY)
gray_norm = cv2.normalize(gray_resized, None, 0, 255, cv2.NORM_MINMAX)

# Auto select brightest pixel as seed

min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(gray_norm)
seed_point = (max_loc[1], max_loc[0]) # (row, col)

region = flood(gray_norm, seed_point, tolerance=50) # 💥 high tolerance

# ----------------------------
# 4. Show Results
# ----------------------------
plt.figure(figsize=(15, 5))

# Original
plt.subplot(1, 3, 1)
plt.imshow(cv2.cvtColor(resized_image, cv2.COLOR_BGR2RGB))
plt.title('Original Image')
plt.axis('off')

# Region Growing
plt.subplot(1, 3, 2)
plt.imshow(region, cmap='gray')
plt.title(f'Region Growing\nAuto Seed: {seed_point}')
plt.axis('off')

# Felzenszwalb
plt.subplot(1, 3, 3)
plt.imshow(felz_segments, cmap='nipy_spectral')
plt.title("Felzenszwalb's Segmentation")
plt.axis('off')

plt.tight_layout()
plt.show()

https://colab.research.google.com/drive/1c-YnbIsIZJ8J0JYjSYZDOKnjdcUwwpVs#scrollTo=KOkv9qjbdGBc&printMode=true 5/9
4/16/25, 10:08 PM Untitled12.ipynb - Colab

import cv2
import numpy as np
import matplotlib.pyplot as plt
from skimage.segmentation import flood

# ----------------------------
# 1. Load and Resize Image
# ----------------------------
image_path = 'br.jpg' # Replace with your image file name
image_color = cv2.imread(image_path)
resized_image = cv2.resize(image_color, (256, 256))
gray_original = cv2.cvtColor(resized_image, cv2.COLOR_BGR2GRAY)

# ----------------------------
# 2. Add Gaussian Noise
# ----------------------------
def add_gaussian_noise(img, mean=0, sigma=50):
noise = np.random.normal(mean, sigma, img.shape).astype(np.float32)
noisy = np.clip(img.astype(np.float32) + noise, 0, 255).astype(np.uint8)
return noisy

gray_noisy = add_gaussian_noise(gray_original)

# ----------------------------
# 3. Apply Thresholding
# ----------------------------
_, basic_thresh = cv2.threshold(gray_original, 127, 255, cv2.THRESH_BINARY)

# ----------------------------
# 4. Region Growing Segmentation
# ----------------------------
# Normalize and auto-pick brightest point as seed
gray_norm = cv2.normalize(gray_original, None, 0, 255, cv2.NORM_MINMAX)
_, _, _, max_loc = cv2.minMaxLoc(gray_norm)
seed_point = (max_loc[1], max_loc[0]) # (row, col)
region = flood(gray_norm, seed_point, tolerance=50)

# ----------------------------
# 5. Display Side-by-Side Comparison
# ----------------------------
plt.figure(figsize=(14, 8))

# --- Before/After Noise ---

plt.subplot(2, 2, 1)
plt.imshow(gray_original, cmap='gray')
plt.title("Original (Grayscale)")
plt.axis('off')

https://colab.research.google.com/drive/1c-YnbIsIZJ8J0JYjSYZDOKnjdcUwwpVs#scrollTo=KOkv9qjbdGBc&printMode=true 6/9
4/16/25, 10:08 PM Untitled12.ipynb - Colab
plt.subplot(2, 2, 2)
plt.imshow(gray_noisy, cmap='gray')
plt.title("After Gaussian Noise")
plt.axis('off')

# --- Thresholding vs Region Growing ---

plt.subplot(2, 2, 3)
plt.imshow(basic_thresh, cmap='gray')
plt.title("Thresholding (Binary @127)")
plt.axis('off')

plt.subplot(2, 2, 4)
plt.imshow(region, cmap='gray')
plt.title(f"Region Growing (Seed: {seed_point})")
plt.axis('off')

plt.tight_layout()
plt.show()

import cv2
import numpy as np
import matplotlib.pyplot as plt
from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler

# ----------------------------------------
https://colab.research.google.com/drive/1c-YnbIsIZJ8J0JYjSYZDOKnjdcUwwpVs#scrollTo=KOkv9qjbdGBc&printMode=true 7/9
4/16/25, 10:08 PM Untitled12.ipynb - Colab
# 1. Load & Resize Image
# ----------------------------------------
image_path = 'br.jpg' # 🔁 Replace with your image filename
image_color = cv2.imread(image_path)

if image_color is None:
raise FileNotFoundError("⚠️ Image not found. Please check the path.")

resized_image = cv2.resize(image_color, (256, 256))

gray = cv2.cvtColor(resized_image, cv2.COLOR_BGR2GRAY)
rows, cols = gray.shape

# ----------------------------------------
# 2. Prepare Features (Intensity + Position)
# ----------------------------------------
X = []
for i in range(rows):
for j in range(cols):
intensity = gray[i, j]
X.append([intensity, i, j])

X = np.array(X)

# Normalize features for better clustering

scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

# ----------------------------------------
# 3. Train KMeans (2 clusters: foreground & background)
# ----------------------------------------
kmeans = KMeans(n_clusters=2, random_state=0, n_init=10)
labels = kmeans.fit_predict(X_scaled)

# Reshape labels into image form

segmented_img = labels.reshape(rows, cols)

# ----------------------------------------
# 4. Display Results
# ----------------------------------------
plt.figure(figsize=(10, 5))

plt.subplot(1, 2, 1)
plt.imshow(gray, cmap='gray')
plt.title('Original Grayscale')
plt.axis('off')

plt.subplot(1, 2, 2)
plt.imshow(segmented_img, cmap='gray')
plt.title('KMeans Segmentation (2 Clusters)')
plt.axis('off')

plt.tight_layout()
plt.show()

https://colab.research.google.com/drive/1c-YnbIsIZJ8J0JYjSYZDOKnjdcUwwpVs#scrollTo=KOkv9qjbdGBc&printMode=true 8/9
4/16/25, 10:08 PM Untitled12.ipynb - Colab

https://colab.research.google.com/drive/1c-YnbIsIZJ8J0JYjSYZDOKnjdcUwwpVs#scrollTo=KOkv9qjbdGBc&printMode=true 9/9