martijnvwezel / tu_delft_computer_vision_lab_final_project_2019

TU Delft Computer Vision Lab Final Project 2019 - 3D reconstruction of a model castle and a teddy bear - Code written in Matlab

TU Delft Computer Vision Lab Final Project 2019

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):

1. Feature extraction and matching using the Harris corner detector and SIFT:

2. Filtering outliers by using the 8-point RANSAC algorithm

3. Chaining - Tracking features that are visible from one to another frame

4. Creation and Stitching of point clouds

5. 3D visualization by assigning corresponding RGB values

How to run the code

• 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.