This project was the final project of the Computer Vision course at the Delft University of Technology in 2019. My assigned project partner was @martijnvwezel (Link to his Repo).
The goal of the project was the 3D reconstruction of a model castle as well as a teddy bear by using a sequence of images made from different perspectives around the object. The total project work is divided into five sections: Feature extraction and matching, RANSAC, Chaining, Stitching and the 3D visualization (for details, please have a look to our final report here):
- Feature extraction and matching using the Harris corner detector and SIFT:
- Filtering outliers by using the 8-point RANSAC algorithm
-
Chaining - Tracking features that are visible from one to another frame
-
Creation and Stitching of point clouds
- 3D visualization by assigning corresponding RGB values
- Clone the repository.
- Download and install the latest version of VLFeat for Matlab here (neccessary for SIFT)
- Add the Final_Project_CV folder (and subfolders) as well as the downloaded VLFeat folder (and subfolders) to your path in Matlab.
- Run the script FINAL_Project_CV.m.