Frank Batten College of Engineering and Technology
PhD Electrical & Computer Engineering
This research project is the continuation of an effort by the Vision Lab to understand human gait and motion using special-purpose imaging sensors and novel computer vision algorithms. The project began using a Motion Capture (MoCap) system which measures 3D human skeleton information in real-time by attaching markers to a human subject and viewing the human motion with a set of cameras at different angles. We developed an algorithm to determine the gender of a subject wearing these sensors. The current phase of this project extends this work using a state-of-the-art flash Lidar sensor. This sensor scans the surface of objects and gives a 3D depth map of the object in real-time. We developed a computer vision system that can estimate a human skeleton from Lidar data, which resembles the MoCap data format. Using these computed 3D skeletons we can perform human identification. The poster outlines the current Lidar-based algorithm using a flow chart to explain the input and output for each component of the system and how each data modality is used to build the final skeleton. The poster compares performance for our human identification with other methods for human- identification using 3D sensors. Finally, future work using infrared sensing is discussed.
Biomedical Engineering and Bioengineering | Electrical and Computer Engineering
Download Full Text (887 KB)
Glandon, Alexander; Vidyaratne, Lasitha; Sadeghzadehyazdi, Nasrin; Dhar, Nibir; Familoni, Jide; Acton, Scott; and Iftekharuddin, Khan, "3D LIDAR Human Skeleton Estimation for Long Range Identification" (2021). College of Engineering & Technology (Batten) Posters. 4.