I have 2 different images, one image is a red channel and another image is a green channel. I finally have a merged imaged which shows both red and green channels. I have attached all 3 images. Essentially, I want to find the percent overlap (how much of the green image is being overlapped with the red image. By this, the pixel overlap between red and green).

I was wondering what type of python package I could use or any other helpful advice that I can do this.

Overlay:

Green:

Red:

CodePudding user response：

Here's how you can find the overlap regions of an image:

import os
import cv2
import matplotlib.pyplot as plt
green = cv2.imread('test_images/green_image.jpg')
red = cv2.imread('test_images/red_image.jpg')

plt.imshow(green)

plt.imshow(red)

green_grey = cv2.cvtColor(green, cv2.COLOR_RGB2GRAY)
red_grey = cv2.cvtColor(red, cv2.COLOR_RGB2GRAY)
bitwiseand = cv2.bitwise_and(green_grey, red_grey)
plt.imshow(bitwiseand)

type(bitwiseand)
>>> numpy.ndarray
bitwiseand.shape
>>> (532, 611)

total = green_grey red_grey
total_pixels = total[total>0].shape[0]

matches = bitwiseand[bitwiseand > 0].shape[0]
percentage = (100*matches/total_pixels)
percentage
>>> 12.341323771795391%

So logically:

1. Load images into cv2 numpy arrays
2. Convert to greyscale
3. Use cv2 bitwise_and to find the overlaps
4. Divide the number of values that are > 0 in bitwiseand & > 0 in the added images together
5. that's it!

hope this helps!

CodePudding user response：

If the red and green channels are binarized, it suffices to count the green pixels and for every green pixel tell if there is a red in the other channel. It is an easy matter of scanning the images.

If the channels are not binarized, do binarize them and proceed as above.