What Is This Project?

  This repository is the bottom-up implementation of panorama images. Panorama image, known as image stitching technology, transforms multiple images into one image perspective. For example, One result image consist of the below images(1.jpg ~ 10.jpg).

1.jpg	2.jpg	3.jpg	4.jpg	5.jpg	6.jpg	7.jpg	8.jpg	9.jpg	10.jpg

• This is the final project of the computer vision lecture 2023 fall semester.
• You can see the detailed procedure in final_report.pdf in Korean.

Prerequisite

• opencv
• eigen
• ubuntu22.04

sudo apt install build-essential libopencv-dev libeigen3-dev

How to use?

1. make a build directory

mkdir build

2. build the code

cd build 
cmake ..
cmake --build .

3. execute

unlimt -s unlimited # this command is mendatory to prevent stack-overflow error
show_img.o

Principle of Panorama

Homograpy transform, a.k.a. perspective transform, transforms a perspective from domain image to target image. Homography transform only works in two certain conditions. The first assumption is that the image's feature comes from a flat plane. The second assumption is that the movement of the image should be pure rotation. In other words, they should not be translated.

Flow chart

• It cover

  • RANSAC (Random Sample Consensus)
  • Stitching algorithm
  • Tranform the image
  • ORB feature implementation (TBD)

• It doesn't cover (use opencv library)

  • image decode module
  • SIFT