Digital Image Process Final Project

My Revision

The images I need to stitch are actually videos of an office took from 4 cameras with different aspects, and I need to stitch these four videos into one. Since the 4 cameras are static, in my opinion, I only have to grab out their first frames, stitch them, then extract the stitching information, and apply to all the video. See the videos here: lab1, lab2,lab3 and lab4. And the images of their first frame are shown below.
For original tutorial and instructions, see this markdown file and this jupyter notebook
The new jupyter notebook file for my case is here. I just copy the original procedure, change the images and parameters. For the images I need to stitch, the feature detector 'orb' doesn't work, so I have to change it to 'brisk'.
Since I don't need the process after 'Blending' block, I didn't run it or check it, also maybe remove the resources it needs accidentally.
The final code is here

There are some other codes in the folderpreprocessing. The most important one is the undistort function, mainly for the most left camera fixing. That image after undistortion is

Name		Name	Last commit message	Last commit date
Latest commit History 13 Commits
docs		docs
imgs		imgs
pre_processing		pre_processing
.gitignore		.gitignore
README.md		README.md
README_original.md		README_original.md
Stitching Tutorial.ipynb		Stitching Tutorial.ipynb
Stitching_Tutorial_Mine.ipynb		Stitching_Tutorial_Mine.ipynb
all4imgs.png		all4imgs.png
labc1_2.jpg		labc1_2.jpg
labc2_2.jpg		labc2_2.jpg
labc3_2.jpg		labc3_2.jpg
labc4_2.jpg		labc4_2.jpg
requirements.txt		requirements.txt
runMe.py		runMe.py
testolabc1.avi		testolabc1.avi
testolabc2.avi		testolabc2.avi
testolabc3.avi		testolabc3.avi
testolabc4.avi		testolabc4.avi