Skeleton Extraction From Volumetric data

This is a research project whose goal was to produce a new method of acquiring the position of human joints from volumetric data extracted using the Kinect depth sensor.

The SEFV project was developed in supervision-collaboration with 3 excellent researchers in the Center of Research and Technology in Thessaloniki. The SEFV algorithm I developed makes use of Volumetric thinning, the minimum spanning tree algorithm and different graph metrics combined to assess which branches are good candidates for possible joint positions.

This was my first professional and research project after I finished my studies in Greece. It was an super-educative experience where I learned a ton of things about 3D and programming.

During this project, I worked with Unity3D for the first time, as well as coded in C++. I also used a bit of Matlab and learned about some cool algorithms like Minimum spanning trees and Kalman filtering.