1. Comparison of two eye tracking devices used on printed images Barbora Komínková The Norwegian Color Research Laboratory Faculty of Computer Science and Media Technology Gjøvik University College, Gjøvik, Norway University of Pardubice, Czech Republic b.kominkova@seznam.cz, http://www.colorlab.no, Gjøvik University College, 7.6.2007,

2. 2 Introduction Aim Experiment Media Images Description Experimental results Discussion and results What the next? Content of presentation

3. 3 Introduction Eye tracking technologies are using in extensive area: usability studies color imaging (image difference) reading studies development of internet pages, etc. To do this kind of studies we need the eye tracker to be precise. Two eye tracking camera have been disposable (Remote eye tracking device and Head-mounted eye tracking device). It was not found that the head mounted eye tracking device was used in printed images.

4. 4 Aim Compare two eye tracking devices used on printed images Head-mounted Eye Tracking Device (HED) and Remote Eye Tracking Device (RED). 1.Is possibility to use the HED on printed images as RED? 2.How to register a data from the HED to the real world coordinates. 3.Find out precisions in the different directions of both devices. (precision in different places, precision in the time aspect, precision on the edges of the image, then stability of the calibration and etc.) 4.Investigate advantages and disadvantages of both devices.

5. 5 Experimental setup - Media Head-mounted eye tracking device (HED) Remote eye tracking device (RED)

6. 6 iView X System Experimental setup - Media

7. 7 Experimental setup - Images AB CD Sequence: A  B  A  C  A  D

8. 8 Two parts of the experiment Experiment with the remote eye tracking device Experiment with the head-mounted eye tracking device. Number of observers – 20 observers  cca 40 data sets from all observers Determination of a dominant eye Instruction – before and during the experiment Calibration Experiment Questionnaire Experiment - description

9. 9 Remote eye tracking device Recording of the eye movement – data file  coordinates of the point of regard (gaze data) BeGaze software Matlab program, Excel program Head-mounted eye tracking device Recording of the eye movement – video (MPEG-2) Converting of the video Stabilization of the video – Simulink model Findings real-world coordinates Matlab program Analysis

10. 10 Distance of points from centre of cross – image A in 1., 3., 5. sequence Blue = Mean Yellow = Median Green = Max Red = Min 1. 3. 5. precision in different places precision in the time aspect Experimental - results Image A Results for the RED:

11. 11 Technical detail according to iView X System: Gaze Position Accuracy 0.5 - 1 deg. Image B Experimental - results

12. 12 MEANMEDIANMAXMIN 31,196229,0273159,63000,6007 Image C Image B Experimental - results

13. 13 The head-mounted eye tracking device is not evaluated at this time The evaluation is in-process Problems at this time: time-consuming stabilization Experimental - results Results for the HED:

14. 14 If real-world coordinates are fixated, I will say that HED can be more precise at least on the same level as RED HED allows the subject to move more freely, which I considered an important issue. But from the questionnaire: Most of the observers felt more comfortable with RED, because: HED - too big/small, observer has to wear it, does not feel freely, etc. The rest of observer: HED – possibility move by head, HED is part of the observer, position was better for look at the picture, focus and concentration was better Discussion and results

15. 15 Continue with evaluation of the remote eye tracking device Complete analysis and evaluation of the head-mounted eye tracking device Statistical evaluation of both eye tracking devices Investigate advantages and disadvantages of both eye tracking devices What the next?

16. 16 THANK YOU FOR YOUR ATTENTION