An FPGA based Real-Time Tracking System for Indoor Environments
Open Access
- Author:
- Sampath Kumar, Vikram
- Graduate Program:
- Electrical Engineering
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- None
- Committee Members:
- Vijaykrishnan Narayanan, Thesis Advisor/Co-Advisor
Vijaykrishnan Narayanan, Thesis Advisor/Co-Advisor
Suman Datta, Thesis Advisor/Co-Advisor - Keywords:
- tracking
image processing
connected components - Abstract:
- Advanced Human Computer Interfacing has become an active field of research in a very short span since its inception. Computing systems need to be built in such a way that they can interact with humans at human interaction speeds. Face Detection and Pedestrian Tracking are few such applications that highlight this fact. A real-time tracking system must be able to track the movement of a person or multiple persons within the field of view of a video camera. The system must exhibit a level of robustness that maintains accuracy and performance in dynamic environments that include multiple occlusions, changing scenery, light intensity variations, and shadows. Additionally, tracking systems may extract details such as shirt/face color, age, and height of persons that have been successfully tracked over a number of frames. The goal is to perform all such tasks in real time at frame rates of at least 30 frames per second. Moreover, small form factor is often a requirement in applications in which the image/video processing must be performed on-site in a non-obtrusive fashion. Smart Cameras are embedded with small FPGAs that can process images on-the-go. For instance, such a system finds great use in building home-care systems for monitoring elderly people in their homes when they are alone. The response time and the size of the system are very vital. This thesis describes an FPGA implementation of a tracking system that gives excellent performance benefits when compared to analogous software implementations on a general purpose processor. The system consists of streaming image processing components which are ideally suited for tracking applications.