Edge detection vs. contour detection in computer vision

Computer vision is a fascinating field that enables machines to interpret and process visual information, allowing them to perform complex tasks like object detection, image recognition, and much more.

Two main techniques used in computer vision for feature extraction are edge detection and contour detection. In this Answer, we'll explore these concepts, their implementation, and their differences.

Edge detection

Simply put, edge detection is a technique used to identify the edges found in an image. We can define an edge to be a change in pixel or image intensity.

Numerous algorithms exist for edge detection. One of the most widely used ones is known as Canny edge detection, which is a multi-step edge detection technique that can detect edges by following these steps:

Gaussian smoothing
Calculating the gradient intensity
Non-maximum suppression
Double thresholding
Edge tracking

Note: Other edge detection algorithms include Sobel, Fuzzy logic, etc.

Edge detection implementation

import cv2
import matplotlib.pyplot as plt

def perform_canny_edge_detection(image_path, min_threshold, max_threshold):
    img = cv2.imread(image_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    edges = cv2.Canny(gray, min_threshold, max_threshold)
    return img, gray, edges

if __name__ == "__main__":
    input_image_path = 'sample.png'
    min_threshold = 50
    max_threshold = 150

    original_img, grayscale_img, edges = perform_canny_edge_detection(input_image_path, min_threshold, max_threshold)

    plt.figure(figsize=(10, 6))

    plt.subplot(1, 3, 1)
    plt.imshow(cv2.cvtColor(original_img, cv2.COLOR_BGR2RGB))
    plt.title('Original Image')

    plt.subplot(1, 3, 2)
    plt.imshow(grayscale_img, cmap='gray')
    plt.title('Grayscale Image')

    plt.subplot(1, 3, 3)
    plt.imshow(edges, cmap='gray')
    plt.title('Edge Detection')

    plt.tight_layout()
    plt.show()

import cv2

def perform_contour_detection(image_path):
    img = cv2.imread(image_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    _, thresh = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY)
    contours, _ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    return contours

if __name__ == "__main__":
    input_image_path = 'sample.png'

    contours = perform_contour_detection(input_image_path)

    img_with_contours = cv2.imread(input_image_path)
    cv2.drawContours(img_with_contours, contours, -1, (0, 255, 0), 2)

    cv2.imshow("Contour Detection", img_with_contours)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

import cv2
import matplotlib.pyplot as plt

def perform_contour_detection(image_path):
    img = cv2.imread(image_path)
    gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    _, thresh = cv2.threshold(gray_img, 127, 255, cv2.THRESH_BINARY)
    contours, _ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    return img, gray_img, contours

if __name__ == "__main__":
    input_image_path = 'sample.png'

    original_img, gray_img, contours = perform_contour_detection(input_image_path)

    img_with_contours = original_img.copy()
    cv2.drawContours(img_with_contours, contours, -1, (0, 255, 0), 2)

    plt.figure(figsize=(15, 5))

    plt.subplot(1, 3, 1)
    plt.imshow(cv2.cvtColor(original_img, cv2.COLOR_BGR2RGB))
    plt.title('Original Image')
    plt.axis('on')

    plt.subplot(1, 3, 2)
    plt.imshow(gray_img, cmap='gray')
    plt.title('Grayscale Image')
    plt.axis('on')

    plt.subplot(1, 3, 3)
    plt.imshow(cv2.cvtColor(img_with_contours, cv2.COLOR_BGR2RGB))
    plt.title('Contour Detection')
    plt.axis('on')

    plt.tight_layout()
    plt.show()

Applications of these techniques

Object detection

We can efficiently detect objects both within images or videos, and we can use edge and contour detection for the feature detection needed in object detection.

Image segmentation

Contours can help segment an image into different meaningful regions. These regions can then be analyzed according to the specific need.

Robotics

We can employ edge detection in different obstacle-detection tasks. For instance, it can be useful in robots assisting the elderly or autonomous vehicles.

Medical image

A great use of contours is in detecting and analyzing objects in medical images, such as tumors or blood vessels.

Condensed differences table

Although both detection mechanisms ultimately find the boundaries of an image, the way they do it is different. Edge detection focuses on simply the drastic changes in brightness or pixels, while contour detection focuses on outlining closed objects, thus producing connected loops each time.

Below is a list of condensed differences between both of these techniques. Choosing which one to apply is a subjective matter and depends on the problem at hand.

Aspect	Edge detection	Contour detection
Robustness	Sensitive to noise and could produce false positives.	More robust to noise and provides accurate boundary representation.
Image type	Applied on grayscale images.	Applied on binary or grayscale images.
Processing	Doesn't usually require post-processing.	Contours may require additional post-processing (e.g., filtering, approximation).
Complexity	Generally simpler and faster to compute.	Can be more complex, especially with complex shapes.
Algorithm	Canny, Sobel, etc. OpenCV's cv. Canny can be used.	OpenCV's cv2.findContours() can be used.
Information	Provides information about intensity gradients.	Provides information about connected boundary points.
Overlapping	Edges may not be closed or connected.	Contours are always closed and connected loops.
Object Holes	Edges sometimes may not enclose object holes.	Contours will enclose object holes.

Edge detection vs. contour detection in computer vision

Feature extraction