How to Approximate a Contour Shape in an Image Using OpenCV Python

Contour approximation is a powerful technique in computer vision and image processing that allows us to simplify complex shapes and boundaries. Essentially, it involves reducing the number of points that define a shape while preserving important features of the original shape. OpenCV is a popular library for computer vision and image processing that provides several algorithms for contour approximation in Python.

In this tutorial, we will be examining how to approximate a contour shape in an image using OpenCV in Python. We will also examine various techniques for shape simplification and smoothing that are commonly used in computer vision.

Prerequisites

Before we begin, make sure that you have OpenCV and NumPy installed in your Python environment. You can use pip to install these packages:

pip install opencv-python
pip install numpy

Image Processing Basics

Before we get into contour approximation, let’s have a brief introduction to image processing basics. Image processing is a field of computer vision that involves extracting important information from images. There are several techniques for image processing such as edge detection, thresholding, morphological operations and contour detection.

In this tutorial, we will be using contour detection to identify the boundaries of an object in an image. Contours are simply the boundaries of a shape or object in an image. Once we have identified contour lines, we can use contour approximation techniques to simplify and smooth the contours.

Contour Detection using OpenCV

First, let’s import the necessary libraries and load the image that we want to process:

import cv2
import numpy as np

img = cv2.imread('shape.jpg')
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

Here, we have imported OpenCV and NumPy, loaded an image called shape.jpg using the cv2.imread() function, and converted the image to grayscale using cv2.cvtColor().

Next, let’s apply edge detection using the Canny edge detection algorithm:

edges = cv2.Canny(gray, 100, 200)

The Canny edge detection algorithm is a popular technique for detecting edges in an image. It works by applying a series of image filters to identify edges in the image.

Now, let’s identify the contours in the image using the findContours() function:

contours, hierarchy = cv2.findContours(edges, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

The findContours() function takes in the input image (in this case, our edges image), a contour retrieval mode, and a contour approximation method. It returns the contours in the image as a list of points.

The contour retrieval mode specifies how we want to retrieve the contours. In this case, we have specified cv2.RETR_EXTERNAL, which means that we want to retrieve only the outer contours of the shapes in the image.

The contour approximation method specifies how we want to approximate the contours. In the next section, we will examine various contour approximation techniques. For now, let’s move on to visualizing the contours in the original image.

Visualizing Contours

We can easily visualize the contours in the original image using the drawContours() function:

cv2.drawContours(img, contours, -1, (0, 255, 0), 2)
cv2.imshow('contours', img)
cv2.waitKey(0)

Here, we have used the drawContours() function to draw the contours in the original image. The function takes in the original image, the list of contours, the index of the contour to draw (-1 to draw all contours), the color of the contour (in this case, green), and the thickness of the contour lines.

Finally, we have used cv2.imshow() to display the image and cv2.waitKey() to wait for a key press before closing the window.

Contour Approximation Techniques

Now that we have identified the contours in our image, let’s examine various techniques for contour approximation.

Ramer-Douglas-Peucker Algorithm

The Ramer-Douglas-Peucker (RDP) algorithm is a popular technique for contour approximation. It works by recursively dividing a contour into smaller line segments until a desired level of approximation is achieved.

We can use the cv2.approxPolyDP() function in OpenCV to apply the RDP algorithm to our contours:

epsilon = 0.01 * cv2.arcLength(contours[0], True)
approx = cv2.approxPolyDP(contours[0], epsilon, True)

cv2.drawContours(img, [approx], -1, (0, 255, 0), 2)
cv2.imshow('approx', img)
cv2.waitKey(0)

Here, we have specified an approximate contour using the cv2.approxPolyDP() function. Thefunction takes in the original contour, the maximum distance from the contour to the approximated shape (in this case, a percentage of the contour perimeter), and a boolean flag that specifies whether the approximated contour is a closed shape.

Convex Hull Algorithm

Another popular contour approximation technique is the Convex Hull algorithm, which simplifies a contour so that it is a convex polygon that completely encloses the original contour.

We can use the cv2.convexHull() function in OpenCV to apply the Convex Hull algorithm to our contours:

hull = cv2.convexHull(contours[0])
cv2.drawContours(img, [hull], -1, (0, 255, 0), 2)
cv2.imshow('hull', img)
cv2.waitKey(0)

Here, we have specified a convex hull using the cv2.convexHull() function. The function takes in a contour and returns the points that define the convex shape that encloses the original contour.

Minimum Enclosing Circle Algorithm

The Minimum Enclosing Circle algorithm approximates a contour with the smallest possible circle that can enclose the shape.

We can use the cv2.minEnclosingCircle() function in OpenCV to apply the Minimum Enclosing Circle algorithm to our contours:

(x, y), radius = cv2.minEnclosingCircle(contours[0])
center = (int(x), int(y))
radius = int(radius)
cv2.circle(img, center, radius, (0, 255, 0), 2)
cv2.imshow('minimum enclosing circle', img)
cv2.waitKey(0)

Here, we have used the cv2.minEnclosingCircle() function to identify the center and radius of the smallest circle that can enclose the shape defined by our contour. We have then drawn this circle in our original image using the cv2.circle() function.

Conclusion

In this tutorial, we have covered the basics of contour approximation in image processing using OpenCV and Python. We have covered various techniques for simplifying and smoothing contours, including the Ramer-Douglas-Peucker algorithm, Convex Hull algorithm, and Minimum Enclosing Circle algorithm.

Contour approximation is a key technique in computer vision and image processing, and is used in a wide variety of applications such as object recognition, shape analysis, and robotics. With the techniques covered in this tutorial, you have a solid foundation for working with contour approximation in your own image processing projects.