Contour Detection

A contour is a curve joining (continuous) points along a boundary. In simple terms, a contour is an outline of an object, shape, irregular figure, curved areas etc. Contour Detection is useful in areas like Object Detection and Recognition, and Shape Analysis.

Syntax

To find contours in an image we use the following syntax :-

cv2.findContours(image, mode, method, contours=None, hierarchy=None, offset=None)

The method, findContours() requires three parameters.

• image specifies the image we want to find the contours in. Using binary images gives us a more accurate output, and hence thesholding is applied to the original image.

• mode specifies the hierarchy that the contours are returned in (relationship between the contours).

  RETR_EXTERNAL gives the outer contours. If contours are enclosed in others, the outermost one is given.

  RETR_LIST gives all contours and doesn't consider any hierarchy between the contours.

  RETR_TREE gives the complete hierarchy of the contours.

  RETR_CCOMP extracts the contours and organizes them into a 2-level hierarchy(outer and inner contours).

• method specifies how to save the points representing the boundary of a contour.

  cv2.CHAIN_APPROX_NONE saves all the points on the boundary of the contour.

  cv2.CHAIN_APPROX_SIMPLE removes redundant points and compresses the contour.

  It's preferred to use CHAIN_APPROX_SIMPLE to save memory. Instead of considering all points of a line, the two extreme points are enough to draw it. Or in the case of a square, the four courners are enough to draw the square. This is how CHAIN_APPROX_SIMPLE works.

The function returns two values, contours and hierarchy. Hierarchy gives us information about the image topology.

Contours is a python list of all the contours in the image. Each contour is a numpy array of (x,y) coordinates of the boundary points of the detected contour.

To view the contours detected, we use the following syntax :-

cv2.drawContours(image, contours, contourIndex, color, thickness=None,)

The method, drawContours() requires the following parameters

• image specifies the source image we want to draw the contours on.
• contours specifies the contours detected on the image.
• contourIndex specifies the index till/the number of contours that are desired to be drawn.
  • -1 can be given to draw all contours.
• color specifies the color of the contour.
• thickness specifies the thickness of the contour

Working and Implementation

Step 1. The image is first converted to grayscale in order to apply thresholding to it. The image after grayscaling would look like this;

Step 2. Simple Thresholding is applied to the image. Applying Thresholding methods would give us a binary image, using which we could expect a better accuracy in detecting contours. The image after applying simple thresholding would look like this;

Step 3. Contours are detecting with the help of cv2.findContours() method.

Step 4. The detected contours can be viewed with the help of cv2.drawContours().

Written by Soumya Vemuri