5 Best Ways to Find the Bounding Rectangle of an Image Contour in OpenCV Python

by

💡 Problem Formulation: In image processing, it’s often necessary to identify and encapsulate contours within bounding rectangles. OpenCV Python offers several methods to accomplish this. Given a binary or edge-detected image, we want to locate contours and create bounding boxes around them to extract or analyze these isolated regions.

Method 1: Using cv2.boundingRect()

OpenCV’s cv2.boundingRect() function calculates and returns the coordinates of the smallest rectangle that contains the entire contour. This rectangle is not necessarily upright and will be aligned with the x and y axis of the image.

Here’s an example:

import cv2

# Assuming 'contours' is a list of contours found using cv2.findContours()
for contour in contours:
    x, y, w, h = cv2.boundingRect(contour)
    cv2.rectangle(img, (x, y), (x+w, y+h), (0, 255, 0), 2)

Output: An image with green rectangles drawn around each contour.

This snippet iterates over each contour, computes the bounding rectangle coordinates, and then draws a green rectangle over the original image using these coordinates. It’s a straightforward way to visualize contours directly aligned with the axis.

Method 2: Using Minimum Area Rectangle with cv2.minAreaRect() and cv2.boxPoints()

The method combining cv2.minAreaRect() and cv2.boxPoints() identifies the minimum area rotated rectangle that encapsulates the contour. This is useful for non-axis aligned contours.

Here’s an example:

import cv2

# Assuming 'contours' was retrieved by cv2.findContours()
for contour in contours:
    rect = cv2.minAreaRect(contour)
    box = cv2.boxPoints(rect)
    box = np.int0(box)
    cv2.drawContours(img, [box], 0, (0, 0, 255), 2)

Output: An image with red, possibly rotated rectangles drawn around each contour.

This code snippet finds a rotated rectangle of the minimal area enclosing the contour and draws it with red lines. The cv2.boxPoints() function converts the rectangle into a box which is a set of points that OpenCV’s drawContours() method accepts.

Method 3: Using Enclosing Circle with cv2.minEnclosingCircle()

For applications requiring a bounding shape rather than a rectangle, cv2.minEnclosingCircle() finds the smallest circle that completely covers the contour. Although not a rectangle, it is a useful alternative for certain tasks.

Here’s an example:

import cv2

# Assuming 'contours' is obtained by cv2.findContours()
for contour in contours:
    (x, y), radius = cv2.minEnclosingCircle(contour)
    center = (int(x), int(y))
    radius = int(radius)
    cv2.circle(img, center, radius, (255, 0, 0), 2)

Output: An image with blue circles drawn around each contour.

The snippet calculates and then draws a blue circle that fully contains the contour. This could be optimal for round-shaped contours where a bounding rectangle would be excessive.

Method 4: Using Fitted Ellipse with cv2.fitEllipse()

When contours are elliptical, using cv2.fitEllipse() could be the most accurate choice. It fits an ellipse to the contour which contains all its points.

Here’s an example:

import cv2

# Assuming 'contours' was found already
for contour in contours:
    if len(contour) >= 5: # A minimum of 5 points is required
        ellipse = cv2.fitEllipse(contour)
        cv2.ellipse(img, ellipse, (0, 255, 255), 2)

Output: An image with cyan ellipses drawn around qualifying contours.

This code snippet cater to situations where the contour is more naturally encapsulated by an ellipse. It avoids the straight-edge bias of rectangles, but it requires a minimum number of points to calculate the ellipse.

Bonus One-Liner Method 5: Using cv2.contourArea() and cv2.arcLength()

As a quick one-liner, cv2.contourArea() and cv2.arcLength() can provide immediate insights into the size and perimeter of the contour, which can be helpful for simple analysis or filtering.

Here’s an example:

import cv2

# Assuming 'contour' is a single contour
area = cv2.contourArea(contour)
perimeter = cv2.arcLength(contour, True)

Output: The area and the perimeter of the given contour.

While this method does not visualize a bounding shape, it offers quick numerical characteristics of the contour, aiding in tasks such as filtering or preliminary analysis.

Summary/Discussion

  • Method 1: cv2.boundingRect(). Fast and straightforward. Aligns with image axes. Not suitable for rotated objects.
  • Method 2: cv2.minAreaRect() and cv2.boxPoints(). Handles rotated contours. More complex than axis-aligned methods. Could be overkill for simple shapes.
  • Method 3: cv2.minEnclosingCircle(). Ideal for circular contours. Ignores shape edges, potentially missing details.
  • Method 4: cv2.fitEllipse(). Perfect for elliptical shapes. Requires a minimum of 5 points. Not suitable for simple or non-elliptical contours.
  • Bonus Method 5: cv2.contourArea() and cv2.arcLength(). Provides fast calculations. Best for filtering or analysis rather than visual bounding.