How to stitch images by openCV with masks? (Only stitch parts of images)

Question

I was trying to stitch images with OpenCV stitcher.

Some of the feed in images have some black areas left by previous processing and they cause some trouble while stitching.

First image

Second image

Third image

Forth image

And my code is like this:

import numpy as np
import cv2
import os
import glob
import argparse

parser = argparse.ArgumentParser(description='Stitch several images.')

parser.add_argument("-i","--inputfiles", type = str, help = "the input files", nargs='+')
parser.add_argument("-o","--outputfiles", type = str, help = "the output files")

args = parser.parse_args()

print "input: " + str(args.inputfiles)
print "potential output: " + args.outputfiles

stitcher = cv2.createStitcherScans(True)
arrayOfImage = []    
for imageName in args.inputfiles:
    arrayOfImage.append(cv2.imread(imageName))
stitchingResult = stitcher.stitch(arrayOfImage)    
cv2.imwrite(args.outputfiles,stitchingResult[1])

And I use command

python stitchingImages.py -i 1.png 2.png 3.png 4.png -o result.png

to test it.

The result is Result

I think the problem is the black areas are recognised as strong features. Is there any way to exclude them from the feature detection and blending?