Python 3 OpenCV Color Object Detect with defined edges

Question

My python code takes a screenshot of by desktop and looks for a red rectangle on a black background, when I use the cv2.findcountours it doesn't return an exact sized rectangle, it seems to be distorted. I would like to obtain the exact area of the shape. Also, the image on my screenshot has no pixelation and borders are sharp. Thanks for your help!

frame_TS_new_raw = np.array(sct.grab(monitor_TS_new))
frame_TS_new = cv2.cvtColor(frame_TS_new_raw, cv2.COLOR_RGBA2RGB)

green_mask_TS_new = cv2.inRange(frame_TS_new,green_lower_range, green_upper_range)

# find contours for New TS
cnts_green_new = cv2.findContours(green_mask_TS_new.copy(), cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)
cnts_green_new = cnts_green_new[0] if imutils.is_cv2() else cnts_green_new[1]
if len(cnts_green_new) > 0:
    for c in cnts_green_new:
        # if the contour is not sufficiently large, ignore it
        if cv2.contourArea(c) > 100:
            area = cv2.contourArea(c)

screenshot of the masked and unmasked

The image on the left is raw screenshot and the image on the right is the masked.

What do you mean "image on the left" and "image on the right" -- there's just one giant white image with two tiny blobs in the top left corner... — Dan Mašek
– Dan Mašek, Commented Apr 24, 2018 at 18:04

user8190410 · Accepted Answer · 2018-04-24 19:24:45Z

1

To find area of red rectangle in the image you can do as:

extract red channel of the image
threshold red channel
count number of nonzero pixels

Example code:

import cv2
import numpy as np

img = cv2.imread('vju0v.png')
img = np.array(img) # convert to numpy array
red = img[:,:,2]    # extract red channel
rect = np.float32(red > 200)   # find red pixels
area = np.sum(rect)    # count nonzero pixels

print('Area = ' + str(area))

cv2.imshow('Red Channel',rect)
cv2.waitKey(0)

answered Apr 24, 2018 at 19:24

user8190410

1,3142 gold badges8 silver badges17 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

mlewk Over a year ago

Is there any way that I can use this method of counting the pixels but use a range because I need to detect three different colors of red but I dont want to overlap other colors (dark red and gray for example)?

user8190410 Over a year ago

you can define your range as rect = np.float32(red > 200 and red < 225)

mlewk Over a year ago

What about a range for the other colors like if I want red to be from 30 to 255, green 0 to 100, blue 0 to 85?

user8190410 Over a year ago

you can get three channels as red = img[:,:,2] and green = img[:,:,1] and blue = img[:,:,0] and use your range to get colors

mlewk Over a year ago

I get this error: rect = np.float32(red > 30 and red < 255) # find red pixels ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

Collectives™ on Stack Overflow

Python 3 OpenCV Color Object Detect with defined edges

1 Answer 1

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related