Introduction

Today I will talk about how to recognize the contour of the desired color using python / opencv, such a task is often found in robotics, and all kinds of automation.

Using the proposed solution, you can, for example, distinguish the contour of a line beyond which a robot should not travel, or an object for a copter. Such a task may arise if you need to, for example, ~~see that an orange is put in the basket~~ .

What to learn

Initially, I assume that a person has a python ide (practically any will do) and basic knowledge of python.

Install the OpenCV library using the following command:

pip install opencv-python

And numpy:

pip install numpy

The code

Add the necessary libraries:

import cv2
import numpy as np

Create a cup object where the video will go. 0 - default camera, 1 - custom.

cap = cv2.VideoCapture(0)

_, frame = cap.read()

reads a frame. instead of _ there may be a rate - fps cameras will go there, but in this case I do not use it.

And now let's process the frame for further convenience. Reduce:


    frameresized = cv2.resize(frame, None, fx=2, fy=2, interpolation=cv2.INTER_CUBIC)
    hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV) #     BGR2HSV
    hsv = cv2.blur(hsv, (5, 5))  #    
    mask = cv2.inRange(hsv, (89, 124, 73), (255, 255, 255))
    #   
    lower_blue = np.array([38, 86, 0])
#    (   )

Let's take a closer look at the resize function:

frame -   
fx -     
fy -     Y
interpolation - INTER_NEAREST -    
INTER_LINEAR -   (  )
INTER_AREA -       .                  INTER_NEAREST method.
INTER_CUBIC -   4x4  
INTER_LANCZOS4 -    8x8

And now let's recognize the contours on the processed frame and sort them:

 contours, _ = cv2.findContours(mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE)
    counturs = sorted(contours, key=cv2.contourArea, reverse=True)

Now let's look at findContours in more detail:

 cv.CHAIN_APPROX_NONE -     (       )
cv.CHAIN_APPROX_SIMPLE -     -    (  )
cv.RETR_TREE -             
cv.RETR_EXTERNAL  -       
cv.RETR_CCOMP -  2  ,               1 ,     
cv2.RETR_TREE -

Visualization RETR_CCOMP:

Visualization RETR_TREE:

And now we will draw the contours that we found:

or contour in counturs:

        cv2.drawContours(frame, counturs[0], -1, (255, 0, 255), 3)
        cv2.imshow("Counturs", frame)  #    
        #cv2.imshow("Mask", mask)
       # cv2.imshow("ret",ret)
        #cv2.imshow("blur",blurred_frame)
    key = cv2.waitKey(1) #  ESC 
    if key == 27:
          break

And at the touch of a button, we stop reading the camera and close all the windows:

cap.release()
cv2.destroyAllWindows()

Repository

Thanks for reading the article.

I would be grateful if you subscribe to my channel in

my game

Upd - it’s funny that the article got about 8 times less views than expected, but at the same time, the estimated number of “bookmarks” is only twice as many. That's how we live.

Computer vision tasks - finding objects of the right color

Introduction

What to learn

The code

More articles: