|Wearable Video Recording System for Gaze Tracking|
Lab: Computational Perception Laboratory, College of Computing, Georgia Institute of Technology, Atlanta, GA
The main task of gaze tracking is to recognize the place where one is looking at. Gaze tracking is very useful in many disciplines such as psychology, cognitive linguistics and industrial design. Now some commercial gaze tracking systems are available (like fig.3). In the NSF project ”Behavioral Imaging”, wearable sensing device need to be developed for children. The most common gaze tracking method which use infrared reflective feature of iris is not suitable for children, since infrared ray is harmful to children’s eyes. So we need to use optical camera to capture the motion of children’s eyes, then track the motion of the gaze.
In this project my first task is to design the wearable device for children, which can capture and record videos. Cortex-A8 ARM controller was chose as the processor of the system. Hardware platform was designed so USB camera and SD card can be connected to the ARM. Embed Linux was transplanted to the ARM processor. OpenCV computer vision library was move into the Linux system. Using the function provide by OpenCV we can capture, store and process the image real-timely.
This project is just started months before. Considering ARM processor has powerful computing capability, OpenCV library can be easily run under Linux. In the future more USB cameras will be connected to the processor so we can get front-view video and inward-view video at the same time, and gaze tracking algorithm will be design on this system.