import cv2

import numpy as np

import time

cap = cv2.VideoCapture(0)

background_capture = cv2.VideoCapture(r'./a.avi')

counter = -1

while cap.isOpened():

    counter += 1

    start_time_extract_figure = time.time()

    # your code

    # extract your figure

    _, frame = cap.read()

    frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)

    mask = np.zeros(frame.shape[:2], np.uint8)

    bgdModel = np.zeros((1, 65), np.float64)

    fgdModel = np.zeros((1, 65), np.float64)

    rect = (50, 50, 450, 490)

    start_time_t = time.time()

    cv2.grabCut(frame, mask, rect, bgdModel, fgdModel, 1, cv2.GC_INIT_WITH_RECT)

    during_time = time.time() - start_time_t

    print('{}-th t_time: {}'.format(counter, during_time))

    mask2 = np.where((mask == 2) | (mask == 0), (0,), (1,)).astype('uint8')

    frame = frame * mask2[:, :, np.newaxis]

    elapsed_time_extract_figure = time.time() - start_time_extract_figure

    print('{}-th extract_figure_time: {}'.format(counter, elapsed_time_extract_figure))

    # extract the background

    start_time_combination = time.time()

    # your code

    ret, background = background_capture.read()

    background = cv2.resize(background, (640, 480), interpolation=cv2.INTER_AREA)

    # maybe the default size of embedded camera is 640x480

    # combine the figure and background using mask instead of iteration

    mask_1 = frame > 0

    mask_2 = frame <= 0

    combination = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB) * mask_1 + background * mask_2

    elapsed_time_combination = time.time() - start_time_combination

    print('{}-th combination_time: {}'.format(counter, elapsed_time_combination))

    cv2.imshow('combination', combination)

    k = cv2.waitKey(30) & 0xff

    if k == 27:

        break

cap.release()

cv2.destroyAllWindows()

发现影响速度的最主要因素是grabCut函数, 因此附官网grabCut详细说明https://docs.opencv.org/trunk/d8/d83/tutorial_py_grabcut.html供以参考.

• img - Input image
• mask - It is a mask image where we specify which areas are background, foreground or probable background/foreground etc. It is done by the following flags, cv2.GC_BGD, cv2.GC_FGD, cv2.GC_PR_BGD, cv2.GC_PR_FGD, or simply pass 0,1,2,3 to image.
• rect - It is the coordinates of a rectangle which includes the foreground object in the format (x,y,w,h)
• bdgModel, fgdModel - These are arrays used by the algorithm internally. You just create two np.float64 type zero arrays of size (1,65).
• iterCount - Number of iterations the algorithm should run.
• mode - It should be cv2.GC_INIT_WITH_RECT or cv2.GC_INIT_WITH_MASK or combined which decides whether we are drawing rectangle or final touchup strokes.