I have an image stack. I am trying to use multiprocessing to do some process on each image in stack to get new image, and then
bitwise_or with the old image.
Since multiprocessing won't keep the original order, I pass image index as
bitwise_or the new image and old image by a for loop.
Is there a way to avoid for-loop for
bitwise_or to have a better performance? Thanks
import cv2 as cv import numpy as np from multiprocessing import Pool def process_img(idx, img): new_img = do_something(img) return idx, new_img def main() # ... output_stack = np.copy(img_stack) with Pool() as pool: args = zip( range(len(img_stack)), img_stack) new_imgs_with_idx = pool.starmap(process_img, args) for i, new_img in new_imgs_with_idx: # Any better ways? output_stack[i] = cv.bitwise_or(output_stack[i], new_img) # ... return output_stack