This algorithm runs through a video file and stitches together all the frames in the video to form a canvas. It can be used for scanning in large documents where the resolution from a single photo may not be sufficient. Currently this doesnt take into account image blurring, evaluating whether an incoming frame has a better quality than the previous one, or lens undistortion.
Getting the app to run is pretty easy, just follow the script below! This script will not install OpenCV or Numpy
# Clone the repo
git clone https://github.com/WillBrennan/VideoStitcher && cd VideoStitcher
# Run the algorithm!
python main.py <path to video file> --display --save
Automatic Panoramic Image Stitching using Invariant Features