如何获得仅由 0 和 1 组成的 3D 矩阵的 3D 坐标

Question

I have a binarized mask, consisting of only 1s and 0s. I want to get all of the locations in the image for the 0s so that I can compare it to the image background.

I am loading the masks by:

import PIL
from PIL import ImageOps, Image
from tensorflow.keras.preprocessing.image import load_img
from numpy import asarray

mask_path = '/Users/mshah/Downloads/8_24_Test2/Masks/3_Agar_Site_3_cp_masks.png'


loaded_mask = PIL.ImageOps.autocontrast(load_img(mask_path))
mask_array = asarray(loaded_mask)

zero_array = np.where(mask_array == 0)

This gives me a tuple with 2 indexes which I don't understand how to use. shape for mask_array is (160, 160) and dtype is uint8 .

Note: the reason that I am using autocontrast instead of just Image.open is because Image.open doesn't work with binarized PNGs (at least the way I'm using).

Any suggestions/help?

EDIT: I found the answer, it is posted below

Answer 1

I found the answer via How to retrieve, in Python, row and column numbers of the elements in a matrix that match a certain value? .

I created an np.ones array with size (160, 160) and then subtracted the mask image from the ones array to get an array that contained 1s for where the background was.

shape = mask_array.shape

ones_array = np.ones(shape, dtype='uint8')

sub_array = ones_array - mask_array

Then I used list(zip(*np.where(sub_array == 1))) to create a list of where the coordinates are by zipping the x and y's together.