Our code is able to take any picture and replace the sky background of the video. To achieve this goal, we track the video, separate the sky from the rest of the image, and then replace the sky part of the video with an image. In order to separate the sky in each frame, we calculate how likely each pixel is the sky by its brightness and height in each frame. The brighter and higher the pixel is, the more likely it is a pixel of the sky. We then replace each pixel with corresponding pixel in the picture in proportion to its likeness to the sky
src/ - Directory with source code
Makefile - Unix/Mac makefile for building the project with "make".
imagepro.[vcproj/sln/suo] - Project file for Visual Studio 2005 on Windows
imgpro.cpp - Main program, parses the command line arguments, and calls the appropriate image functions
R2Image.[cpp/h] - Image class with processing functions (this is the only file that you need to edit)
R2Pixel.[cpp/h] - Pixel class
R2/ - A library of useful 2D geometric primitives
jpeg/ - A library for reading/writing JPEG files
input/ - Contains example input images.
output/ - Es empty to start -- it will contain the images produced by your program (see below)
runme.bat - a script (for Windows) that you will fill in to demonstrate execution of your program
runme.sh - same as <code>runme.bat, but for Mac OS X
If you are developing on a Windows machine and have Visual Studio installed, use the provided project solution file (assn1.sln) in the src/ directory to build the program.
If you are developing on a Mac or Linux machine, cd into the src/ directory and type "make". In either case, an executable called imgpro (or imgpro.exe) will be created in the src/ directory.
# Clone this repository
$ git clone https://github.com/XinyuYang/VideoProcessing.git
# Go into the repository
$ cd VideoProcessing
# To see all the options of operations
$ src/imgpro -help
# Image processing (Drag the image into the input folder, put the method name in the end of the command line such as brighntess,blur,harris...)
$ src/imgpro input/testpattern.jpg output/testpattern_brighntess_0.5.jpg -brightness 0.5
# Video processing (Drag the images of the video into the videoinput folder; create a new folder called "videooutput" to save the output sequence)
$ src/imgpro -video
FFmpeg
# The FFmpeg command line syntax to convert a video into a JPG sequence
$ ffmpeg -i video.mp4 -qscale:v 1 videoinput/input%07d.jpg
# The FFmpeg command line syntax to convert a JPG sequence into a video (24fps)
$ ffmpeg -r 24 -f image2 -s 1920x1080 -i videooutput/output%07d.jpg -vcodec libx264 -crf 25 -pix_fmt yuv420p test.mp4
Link to all test videos:
https://drive.google.com/drive/folders/1KeugXYiWssy2Q9LSo9bAoQtYl70Z99Oe?usp=sharing
- Xinyu Yang
- Weiqiu You
- Megan Zhao