Constantinos Glynos

3D Skeleton Extraction using a single Kinect camera


Nowadays, it is well established that Computer Vision is used in many applications and devices. One area of huge interest is the detection and reconstruction of the human skeletal structure. In short words, the camera will identify the human parts and construct the joints of the person standing in front of it. So far, scientists have managed to build various systems that reconstruct the human skeleton representing it as lines and spheres in 2D using one camera and 3D using two or more cameras. Although there are still a few areas requiring improvements for such methods, such as computational cost and stability, the results are generally accurate and efficient. Currently, the reconstruction of the human skeleton in 3D using one camera is still under research. In this master thesis a relatively novel approach is proposed, using image-processing techniques, mesh extraction, line skeletonization, 2D Skeleton extraction and depth detection methods, resulting into a 3D skeleton