Change the colors within certain range to another color using OpenCV

Question

I want to change the brown areas to RED (or another color). Just I don't know how to get the ranges for brown and put them in python code. I know how to change a single color, but not a range of colors. Any Ideas? Thanks

Answer 1

This should give you an idea - it is pretty well commented:

#!/usr/local/bin/python3
import cv2 as cv
import numpy as np

# Load the aerial image and convert to HSV colourspace
image = cv.imread("aerial.png")
hsv=cv.cvtColor(image,cv.COLOR_BGR2HSV)

# Define lower and uppper limits of what we call "brown"
brown_lo=np.array([10,0,0])
brown_hi=np.array([20,255,255])

# Mask image to only select browns
mask=cv.inRange(hsv,brown_lo,brown_hi)

# Change image to red where we found brown
image[mask>0]=(0,0,255)

cv.imwrite("result.png",image)

---

How did I determine the limits for "brown"? I located a brown area in the image, and cropped it out to remove everything else. Then I resized it to 1x1 to average all the shades of brown in that area and converted it to HSV colourspace, I printed that and took the value for Hue which was 15 and went +/-5 to give a range of 10-20. Increase the range to 8-22 to select a wider range of hues.

HSV/HSL colourspace is described on Wikipedia here.

Keywords: Image processing, Python, OpenCV, inRange, range of colours, prime.

Answer 2

I would like to propose a different approach. However, this will work only for a range of certain dominant colors (red, blue, green and blue). I am focusing on the red colored regions present in the image in question.

Background:

Here I am using LAB color space where:

• L-channel: expresses the brightness in the image
• A-channel: expresses variation of color in the image between red and green
• B-channel: expresses variation of color in the image between yellow and blue

Since I am interested in the red region, I will choose the A-channel for further processing.

Code:

img = cv2.imread('image_path')

# convert to LAB color space
lab = cv2.cvtColor(img, cv2.COLOR_BGR2LAB)

# A-channel
cv2.imshow('A-channel', lab[:,:,1])

If you look at the image closely, the bright regions correspond to the red color in the original image. Now when we threshold it, we can isolate it completely:

th = cv2.threshold(lab[:,:,1],127,255,cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]

Using the th image as mask, we give a different color to the corresponding regions in white:

# create copy of original image
img1=img.copy()
# highlight white region with different color
img1[th==255]=(255,255,0)

Here are both the images stacked beside each other:

You can normalize the A-channel image to better visualize it:

dst = cv2.normalize(lab[:,:,1], dst=None, alpha=0, beta=255,norm_type=cv2.NORM_MINMAX, dtype=cv2.CV_8U)

In this way, there is no need to look for range in HSV space when working with dominant colors. Exploring the B-channel can help isolate blue and yellow colored regions.