How to detect black contour in image using open-cv

Question

I have an image showed below (this image is a result of pre-processing of original image I had):

What I am trying to do is to detect the contour for the Black region bounded by 2 white regions (please note that white regions are not always solid and there could be 2, 3,... disconnected white region on top - for most of pictures the white regions are continuous; however, there are some examples where top white region is not continuous - Please see update section of my question to see example of non-continuous region)

Do you have any suggestion as how I can detect the boundary of black region and separate the black region?

========= UPDATE Requested by Micka ========

Example of non-continuous white region (this is happening only on top layer) - bottom layer is always continues solid white region

In this case like the picture showed above, I am okay with selecting a region like the one I showed below

ideal selection

Original PHOTO I also included the original image for the above pre-processed image below in a case you are interested to work on pre-processing it. The pre-processing code I used and ended up with the image above is showed below:

img = cv2.imread('....tif')
# convert to grayscale
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)

# threshold
thresh = cv2.threshold(gray, 128, 255, cv2.THRESH_BINARY)[1]

# apply morphology
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (1,5))
morph = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, kernel)
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (29,1))
morph = cv2.morphologyEx(morph, cv2.MORPH_CLOSE, kernel)

Answer 1

Here is how:

import cv2
import numpy as np

img = cv2.imread(r"D:\OpenCV Projects\Black Contour\image.jpg")

img_hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
lower = np.array([0, 0, 75])
upper = np.array([179, 179, 115])
mask = cv2.inRange(img_hsv, lower, upper)

mask_blur = cv2.GaussianBlur(mask, (3, 3), 1)
_, thresh = cv2.threshold(mask_blur, 200, 255, cv2.THRESH_BINARY)

contours, _ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)

cv2.drawContours(img, [max(contours, key=cv2.contourArea)], -1, (255, 0, 0), 3)
cv2.imshow("Image", img)
cv2.waitKey(0)
cv2.destroyAllWindows()

Output:

Here is all the processing the image went through during the execution of the above code:

BGR to HSV:

HSV to mask:

Mask to blurred mask:

Blurred mask to thresholded mask:

From here, simply detect the contour from the image above and draw it onto the image.

Answer 2

Here is one approach in Python/OpenCV that I tested on your short lines image. You may have to adjust for other images.

The idea is to apply vertical morphology to close up the spacing. Then get the absdiff between he threshold version and the closed version. The clean up any small spots with a morphology open.

Input:

import cv2
import numpy as np

# read input
image = cv2.imread("short_lines.jpg")

# convert to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

# create a binary thresholded image
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY+cv2.THRESH_OTSU)[1]

# apply morphology close to connect space between lines
kernel = np.ones((45,1), np.uint8)
morph = cv2.morphologyEx(thresh, cv2.MORPH_CLOSE, kernel)

# get absdiff of morph and thresh
diff = cv2.absdiff(thresh, morph)

# apply morphology open to clean up small spots
kernel = np.ones((5,5), np.uint8)
result = cv2.morphologyEx(diff, cv2.MORPH_OPEN, kernel)

# save results
cv2.imwrite("short_lines_close.jpg",morph)
cv2.imwrite("short_lines_diff.jpg",diff)
cv2.imwrite("short_lines_space.jpg",result)

# display result, though it won't show transparency
cv2.imshow("morph", morph)
cv2.imshow("diff", diff)
cv2.imshow("result", result)
cv2.waitKey(0)

Morphology Close image:

Absdiff Image:

Final Result after Morphology Open:

Answer 3

Quick & dirty method:

Scan every image column and count the transitions from black to white and conversely. Every time you find four transitions, the two middle ones give you the thickness. Keep the median value.