Create mask or boundary from each other in Python3

Question

I would like to get the mask from images with boundaries or boundaries from images of masks.

Here are two images:

I went through the followings, I did not understand the logic there: link1, link2, link3, link4

thanks, ilyas

MASK->CONTOUR: use findContours with RETR_EXTERNAL. CONTOUR->MASK: use findContours with RETR_TREE, use internal contours only (use contourArea [see here](https://stackoverflow.com/a/42293993/5008845) ), then use drawContours. Try it out and come back if you have a specific programming problem and some code — Miki, Apr 23 '20 at 10:51
@Miki I am not sure about the c++ code in the link you provided. Could you try with python3 and provide the code? I tried with watershed etc, but I could not. What I want to do: I would like to draw boundaries by hand for any the shapes (cell, nucleus, etc). Then, create a mask from these boundaries. Or instead of drawing boundaries, first I create mask by hand, then draw boundaries. — m.i.cosacak, Apr 23 '20 at 17:01
I found this: [link](https://stackoverflow.com/questions/40441910/fast-method-to-retrieve-contour-mask-from-a-binary-mask-in-python). `im = cv.imread('mask.png')` `im = cv.cvtColor(im,cv.COLOR_BGR2GRAY)` `idx = cv2.findContours(im,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_NONE)[1][0]` `out = np.zeros_like(im)` `out[idx[:,0],idx[:,1]] = 255` — m.i.cosacak, Apr 23 '20 at 22:20
But still cannot create boundaries (second image), but if using gradient function, it gives some boundaries but not correct! — m.i.cosacak, Apr 23 '20 at 22:23
Now this [link] (https://docs.opencv.org/3.0-beta/doc/py_tutorials/py_imgproc/py_contours/py_contours_begin/py_contours_begin.html) helped a lot: — m.i.cosacak, Apr 23 '20 at 22:47

score 0 · Answer 1 · answered Apr 23 '20 at 22:58

from this link one can use the following;

to get boundary from mask:

im = cv2.imread('mask.png')
imgray = cv2.cvtColor(im,cv2.COLOR_BGR2GRAY)
ret,thresh = cv2.threshold(imgray,127,255,0)
contours, hierarchy = cv2.findContours(thresh,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_NONE)
tmp = np.zeros_like(im)
boundary = cv2.drawContours(tmp, contours, -1, (255,255,255), 1)
boundary[boundary > 0] = 255
plt.imsave("mask_boundary.png", boundary, cmap = "gray")

score 0 · Accepted Answer · answered Apr 24 '20 at 21:53

I still do not understand the logic, however, one can create boundary from mask and mask from boundary.

import os
import numpy as np
import cv2
from matplotlib import pyplot as plt

out = "mask_boundary.png"
inp = "mask.png"
im = cv2.imread(inp)
imgray = cv2.cvtColor(im,cv2.COLOR_BGR2GRAY)
ret,thresh = cv2.threshold(imgray,127,255,0)
contours, hierarchy = cv2.findContours(thresh,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_NONE)
tmp = np.zeros_like(im)
boundary = cv2.drawContours(tmp, contours, -1, (255,255,255), 1)
plt.imsave(out, boundary, cmap = "gray")


out = "boundary_mask.png"
inp = "mask_boundary.png"
im = cv2.imread(inp)
imgray = cv2.cvtColor(im,cv2.COLOR_BGR2GRAY)
ret,thresh = cv2.threshold(imgray,0,255,0)
contours, hierarchy = cv2.findContours(thresh,cv2.RETR_TREE,cv2.CHAIN_APPROX_NONE)
tmp = np.zeros_like(im)
boundary = cv2.drawContours(tmp, contours, 0,(255,255,255), -1)
for i in range(1,len(contours)):
    boundary = cv2.drawContours(boundary, contours, i,(255,255,255), -1)

plt.imsave(out, boundary, cmap = "gray")



out = "boun_mask.png"
inp = "boundary.png"
im = cv2.imread(inp)
imgray = cv2.cvtColor(im,cv2.COLOR_BGR2GRAY)
ret,thresh = cv2.threshold(imgray,0,255,0)
contours, hierarchy = cv2.findContours(thresh,cv2.RETR_TREE,cv2.CHAIN_APPROX_NONE)
tmp = np.zeros_like(im)
boundary = cv2.drawContours(tmp, contours, 0,(255,255,255), -1)
for i in range(1,len(contours)):
    boundary = cv2.drawContours(boundary, contours, i,(255,255,255), -1)

plt.imsave(out, boundary, cmap = "gray")

Note: using the mask image above, I first created boundary and then from this boundary I created mask. However, I could not convert all boundary above to mask, especially the one at the image edges/border. It looks like there must be boundary at the border as well! Any help on that will be appreciated!

convert the above mask to boundary:

now convert the new boundary to mask:

now convert the above boundary to mask:

Create mask or boundary from each other in Python3

2 Answers2