A re-implementation of MATLAB code with c++.
A sparse subspace clustering method, and more details can be found here paper MATLAB code.

You need to install eigen3 library for using because the data is calculated and stored in a matrix form.

• Eigen3(Version 3.3.4): Installation tutorial
  
• libigl: a header-only library Homepage.
  

• OpenCV(Version 2.4)
  
• CMake(Version 3.5)
  

cd SCC
cd build
cmake ..
make

You can find the Kmeans algorithm used in the code here.

The result shows that features of pedestrian's area of this image show different motion trajectories, which results in these feature points being classified into the same cluster.