Cartooning an Image using OpenCV - Python

Turning an image into a cartoon-like visualization can be a fun project to understand multiple techniques within computer vision. In this tutorial, we'll transform an image into its cartoon version using OpenCV in Python.

Cartooning an Image using OpenCV

Setup: If you haven't installed OpenCV, do so using pip:
```
pip install opencv-python
```

Read the Image: Start by reading the image.

import cv2

img = cv2.imread("path_to_image.jpg")

Apply a Bilateral Filter: Bilateral filters are great at preserving edges while reducing color noise. We'll use it to give our image a smooth appearance.
```
# Apply bilateral filter with d=9, sigmaColor=sigmaSpace=75
filtered_img = cv2.bilateralFilter(img, d=9, sigmaColor=75, sigmaSpace=75)
```
Convert Image to Grayscale & Apply Median Blur: The idea here is to detect and enhance the edges. So, we first convert the image to grayscale and then apply a median blur to reduce image noise.
```
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
blurred_gray = cv2.medianBlur(gray, 7)
```

Use Adaptive Thresholding: To detect and enhance the edges in the image, we'll use adaptive thresholding.

edges = cv2.adaptiveThreshold(blurred_gray, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, blockSize=9, C=2)

Convert Edges to Color Image: We need the edges in the same format (3-channel) as the filtered image to merge them.
```
color_edges = cv2.cvtColor(edges, cv2.COLOR_GRAY2BGR)
```
Cartoon Effect: Subtract the color edges from the filtered image to give a cartoon effect.
```
cartoon = cv2.bitwise_and(filtered_img, color_edges)
```

Display the Results:

cv2.imshow("Original Image", img)
cv2.imshow("Cartoon Image", cartoon)
cv2.waitKey(0)
cv2.destroyAllWindows()

When you run the entire sequence, you should see two windows: one with the original image and another with the cartoon version of the image.

This process gives a simple cartoon effect. You can experiment with the parameters in the bilateral filter, median blur, and adaptive thresholding to fine-tune the effect according to your needs.

1. Edge Detection:

Edge detection helps identify prominent edges in the image, creating a sketch-like effect.

import cv2
import numpy as np

def edge_detection(image):
    # Convert the image to grayscale
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    
    # Apply GaussianBlur to reduce noise and improve edge detection
    blurred = cv2.GaussianBlur(gray, (5, 5), 0)
    
    # Use Canny edge detector
    edges = cv2.Canny(blurred, 50, 150)
    
    return edges

2. Color Quantization:

Color quantization reduces the number of colors in the image to give it a more cartoonish appearance.

def color_quantization(image, k=9):
    # Reshape the image to a 2D array of pixels
    pixels = image.reshape((-1, 3))
    
    # Convert to float32 for k-means
    pixels = np.float32(pixels)
    
    # Define criteria and apply k-means
    criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 100, 0.2)
    _, labels, centers = cv2.kmeans(pixels, k, None, criteria, 10, cv2.KMEANS_RANDOM_CENTERS)
    
    # Convert back to 8-bit values
    centers = np.uint8(centers)
    
    # Map the labels to the centers
    segmented_image = centers[labels.flatten()]
    
    # Reshape back to the original image shape
    segmented_image = segmented_image.reshape(image.shape)
    
    return segmented_image

3. Combine Edge and Color Quantization:

Combine the edge-detected image and the color-quantized image to create the final cartoon effect.

def cartoonize_image(image):
    # Step 1: Edge Detection
    edges = edge_detection(image)
    
    # Step 2: Color Quantization
    quantized = color_quantization(image)
    
    # Combine the edges and color-quantized image
    cartoon = cv2.bitwise_and(quantized, quantized, mask=edges)
    
    return cartoon

4. Sample Code for Cartoonizing an Image:

Load an image and apply the cartoonize function.

# Load an image
image = cv2.imread('input_image.jpg')

# Cartoonize the image
cartoon_image = cartoonize_image(image)

# Display the original and cartoonized images
cv2.imshow('Original Image', image)
cv2.imshow('Cartoonized Image', cartoon_image)
cv2.waitKey(0)
cv2.destroyAllWindows()

5. Cartoon Filter Parameters and Adjustments:

Experiment with parameters like kernel size, Canny thresholds, and the number of colors (k) in k-means for fine-tuning.

6. Cartooning Techniques with Bilateral Filter:

The bilateral filter can be used for smoothing while preserving edges in the cartoonized image.

def cartoonize_with_bilateral(image):
    # Step 1: Edge Detection
    edges = edge_detection(image)
    
    # Step 2: Apply Bilateral Filter for smoothing
    smooth = cv2.bilateralFilter(image, d=9, sigmaColor=300, sigmaSpace=300)
    
    # Combine the edges and smoothed image
    cartoon = cv2.bitwise_and(smooth, smooth, mask=edges)
    
    return cartoon