5 Best Ways to Cartoonize an Image Using OpenCV in Python

by

💡 Problem Formulation: The goal is to transform a standard digital image into a stylized ‘cartoon’ representation using Python with the aid of the OpenCV library. This involves processes that may include edge detection, color simplification, and smooth shading. Input would be a regular color image, and the desired output is a cartoonish version of that image, with bold contour lines and homogeneous color regions.

Method 1: Applying Bilateral Filtering and Edge Detection

The bilateral filter is used to reduce color palette while preserving edges, followed by edge detection to emphasize outlines. This combination creates the cartoon effect by making the colors more homogenous and the edges crisper, resembling the look of hand-drawn cartoons.

Here’s an example:

import cv2

# Read the image
img = cv2.imread('image.jpg')

# Apply bilateral filter
cartoon_img = cv2.bilateralFilter(img, d=9, sigmaColor=75, sigmaSpace=75)

# Convert to grayscale and apply median blur
gray = cv2.cvtColor(cartoon_img, cv2.COLOR_BGR2GRAY)
blurred = cv2.medianBlur(gray, 7)

# Detect edges and create a mask
edges = cv2.adaptiveThreshold(blurred, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, blockSize=9, C=2)
colored_edges = cv2.bitwise_and(cartoon_img, cartoon_img, mask=edges)

# Show or save the result
cv2.imshow('Cartoonized Image', colored_edges)
cv2.waitKey(0)
cv2.destroyAllWindows()

The output of the above code is an image with smooth color transitions and bold edges, giving it a cartoon-like appearance.

This snippet takes an image and processes it using bilateral filtering for color homogenization and adaptive thresholding for edge detection. Initially, the image undergoes smoothing by the filter that reduces the color palette while maintaining the edge details. Then, grayscale and median blur are applied to highlight only the significant edges. Finally, the adaptiveThreshold function emphasizes these edges to complete the cartoon effect.

Method 2: Color Quantization and Edge Enhancement

Color quantization reduces the number of colors, making the image appear more cartoonish, and edge enhancement makes the outlines more pronounced. This method gives images a distinct flat-color look with emphasized dark lines, akin to classic cartoons.

Here’s an example:

import cv2
import numpy as np

def color_quantization(img, k):
    # Transform the image
    data = np.float32(img).reshape((-1, 3))

    # Determine criteria
    criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 20, 0.001)

    # Implementing K-Means
    ret, label, center = cv2.kmeans(data, k, None, criteria, 10, cv2.KMEANS_RANDOM_CENTERS)
    center = np.uint8(center)
    result = center[label.flatten()]
    result = result.reshape(img.shape)
    return result

# Read the image
img = cv2.imread('image.jpg')

# Color quantization
quantized_img = color_quantization(img, k=9)

# Convert to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# Apply median blur, detect edges
blurred = cv2.medianBlur(gray, 7)
edges = cv2.adaptiveThreshold(blurred, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, blockSize=9, C=2)

# Combine quantized image with edge mask
cartoon_img = cv2.bitwise_and(quantized_img, quantized_img, mask=edges)

# Show or save the result
cv2.imshow('Cartoonized Image', cartoon_img)
cv2.waitKey(0)
cv2.destroyAllWindows()

The result is a visually arresting cartoonized image with a limited color palette and well-defined edges.

In this example, we use K-Means clustering (via cv2.kmeans) for color quantization which clusters pixels into k groups (where k is a small number), thus reducing the color palette. Gaining the quantized image, we then apply median blur and adaptive thresholding for edge detection. The combine step involves masking the quantized image with detected edges, which yields a pure cartoon effect with strong outlines and flat regions of color.

Method 3: Enhancing Edges with a Combination of Filters

This method focuses more on refining and enhancing the edges by utilizing a series of image filters to simulate the sharp ink outlines of cartoons. It’s particularly suitable for images with strong contrast and where you want to emphasize the drawn lines over the color features.

Here’s an example:

import cv2

# Read and scale down the image
img = cv2.imread('image.jpg')
img = cv2.pyrDown(cv2.imread('image.jpg'))

# Convert to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# Apply median blur
blurred = cv2.medianBlur(gray, 5)

# Use a Laplacian filter for edge detection
edges = cv2.Laplacian(blurred, cv2.CV_8U, ksize=5)

# Invert the color of edges
inverted_edges = 255-edges 

# Combine original image with edges
cartoon_img = cv2.bitwise_or(img, img, mask=inverted_edges)
 
# Show or save the result
cv2.imshow('Cartoonized Image', cartoon_img)
cv2.waitKey(0)
cv2.destroyAllWindows()

The code results in a unique cartoon-style image accentuating the edges with a hand-drawn quality.

Here, the focus is on edge detection and enhancement with a Laplacian filter that identifies the transitions in intensity across the image. After blurring, we apply the filter which brings out the edges, then invert the color to convert them into black lines typical for cartoons. Finally, the black lines are superimposed on the original image to create a stark, cartoon-like edge effect.

Bonus One-Liner Method 5: Simplified Cartoon Effect

For a swift, one-liner approach to cartoonizing an image, we can combine OpenCV functionality with a compact code version that delivers a lighter cartoon effect.

Here’s an example:

import cv2

# Reading, smoothing, and edge detection all in one line
cv2.imshow('Cartoonized Image', cv2.bitwise_and(cv2.bilateralFilter(cv2.imread('image.jpg'), d=9, sigmaColor=300, sigmaSpace=300), cv2.bilateralFilter(cv2.imread('image.jpg'), d=9, sigmaColor=300, sigmaSpace=300), mask=cv2.Canny(cv2.imread('image.jpg'), 100, 150)))
cv2.waitKey(0)
cv2.destroyAllWindows()

The one-liner delivers a playful and simplified cartoon effect with minimal hassle.

This streamlined snippet chains several OpenCV functions to produce the cartoon effect in a single line. The image is first read and then fed through a bilateral filter to smooth out the colors while retaining edge detail. The mask for the bitwise_and operation is generated via the Canny edge detector, which picks out the strongest edges. When combined, the result is a quick-and-dirty cartoon filter that can be a starting point for further refinement.

Summary/Discussion

  • Method 1: Bilateral Filtering and Edge Detection. Produces a smooth, clean cartoon effect. Good for high-resolution images but can be computationally intensive.
  • Method 2: Color Quantization and Edge Enhancement. Delivers a visually distinct cartoon with flat colors and crisp borders. Works best with strongly contrasting images. The setup is slightly complex due to k-means.
  • Method 3: Enhancing Edges with a Combination of Filters. Provides a high-contrast edge-focused cartoon. Suitable for stylized images with strong lines. Not ideal for subtle color transitions.
  • Bonus Method 5: Simplified Cartoon Effect. Offers a quick solution with a minimal code footprint. Good for rapid prototyping but lacks the finesse of the other methods.