SFMedu: Structure From Motion for Education Purpose
Version 2 @ 2014
Written by Jianxiong Xiao (MIT License)

This code demonstrates how a traditional structure from motion pipeline is done and how to compute a dense point cloud by matching propagation in a simplest way. It is mainly designed for teaching a computer vision lecture at Princeton. Please check the slides and the homework assignment together with this code release. Although it is for education purpose, it is implemented in a nice way that it actually works for real photos. It is just probably quite slow compared to more optimized system, such as Bundler from Noah Snavely.

Usage: run SFMedu2.m in Matlab.

Compilation:
For the dense matching algorithm, we use a 3rd party Matlab library “Priority Queue” from 
http://www.mathworks.com/matlabcentral/fileexchange/24238-priority-queue--mex-c++-
You may need to recompile the mex files under ./denseMatch/priority_queue_1.0 folder.

Acknowledgement: Thanks Fisher Yu for bug fixing and feedback. We also thanks many people for making their functions publicly available online. Note that the third party library may have different licenses that you need to follow. We include them for the convenience for the users and don't imply any copyright issues.