eye tracking system on embedded platform
play

Eye Tracking System on Embedded Platform Matj ern Department of - PowerPoint PPT Presentation

Eye Tracking System on Embedded Platform Matj ern Department of Electrical Engineering 4.12. 2013 INTRODUCTION Main motivation for this system was the need for high definition image of eye which is fundamental for accurate gaze


  1. Eye Tracking System on Embedded Platform Matěj Černý Department of Electrical Engineering 4.12. 2013

  2. INTRODUCTION Main motivation for this system was the need for • high definition image of eye which is fundamental for accurate gaze tracker. Accurate gaze tracker can be used to control PC with • common GUI by gaze to help disabled people interact with PC or to improve PC work efficiency. High definition image is acquired from PTZ camera • which is controlled by output of Eye Tracker. System should be composed of common components • and should have compact dimensions, for easy use and implementation. Department of Electrical Engineering 4.12. 2013 1

  3. OBJECTIVES Evaluate eye position on embedded platform from • image from standard webcam. Create servo driven mount for high definition camera • with narrow field of view. Create servo driver with USB interface. • Design simple communication protocol between servo • unit and embedded platform. Department of Electrical Engineering 4.12. 2013 2

  4. MATERIAL & METHOD Eye position evaluation software was developed • directly on Pandaboard evaluation board (Cortex A9) in C++ using feature based object detector from OpenCV library. Prototype of servo driven camera mount was built • from two RC servos and building kit Merkur. Servo drive interface was based on Atmel • ATMEGA32U4 providing PWM signals controlling servos and USB interface to communicate with evaluation board. Communication protocol between servo drive and • evaluation board is using virtual COM port on USB and was implemented into both devices. Department of Electrical Engineering 4.12. 2013 3

  5. RESULTS Presented system is very well reflecting expectations • and can be used as a part of accurate gaze tracker. Processing time of one frame is approx. 50ms which is • sufficient for this application but it can be improved using embedded platform with higher performance. Direction accuracy of camera is sufficient and can be • improved by distance measuring device. Department of Electrical Engineering 4.12. 2013 4

  6. ACKNOWLEDGEMENT & CONTACT Matěj Černý matej.cerny @student.upce.cz Department of Electrical Engineering Faculty of Electrical Engineering and Informatics University of Pardubice Czech Republic http://www.upce.cz/en/fei/ke.html The research was supported by the specific research project of the IGA (SGFEI05/2013), University of Pardubice. Department of Electrical Engineering 4.12. 2013 5

Recommend


More recommend