Image Stitching Project

Final project for subject IMP301

Algorithm:

• Extract features using SIFT, SURF, ORB
• Matching features using BruteForce algorithm
• Find homography matrix using RANSAC algorithm
• Blending and warping image

GUI:

• Tkinter library in python

The project is to implement a simple image stitching app, when we input a set of images with overlapped fields, we expect to obtain a wide seamless panorama.

Note

• We have implemented the SIFT algorithm from scratch using Python, however to optimize the execution time in the app we use the API provided by the library.
• Please feel free to check out our implementation for SIFT by the following link SIFT Implementation using Python

Dependency

• Numpy
• Matplotlib
• Pillow

To use all none-free features

• Python 3.7.1
• OpenCV 3.4.2

downgrade opencv version by:

pip install opencv-contrib-python==3.4.2.17 --force-reinstall

To run by newest version (some features doesn't work like SIFT,...)

• Python 3.9
• OpenCV 4.5.3

Usage

• You can directly test project by downloading following .exe file Stitching.exe
• Run main.py or Stitching.exe file and follow the guideline in app by clicking guideline button

Sample (Step by step)

Input images

Matching

Blending

Other examples

Result for 4 pics taken from 5th floor FPT University:

Result for 3 pics taken from classroom at FPT University:

Result for 6 pics taken from Quy Nhon Beach:

Result for 10 pics taken from Quy Nhon city: