EYE MOVEMENT ANALYSIS FOR ACTIVITY RECOGNITION USING ELECTROOCULOGRAPHY

SACHIN MATHEW ECE 09240

RELAVANCE OF THE TECHNOLOGY CURRENT CONFIGURATIONS

Accelerometers or gyroscopes
Reed switches or light sensors Video camera- mobile eye and IVIEW XHED

DISADVANTAGES
Only User

physical activity is sensed intention is completely unexplored

CONTENTS

Introduction Advantages Electrooculography Eye movement types Electrooculogram Architecture for eye based activity recognition Electrode placement Apparatus Performance Human computer Interface Application Conclusion

INTRODUCTION

HUMAN activity recognition -application area for pattern recognition.

The movement patterns of eyes -potential to reveal the activities themselves Instead of sensors here use electrooculography, most advanced form.
Eye movements provide useful information for activity recognition. Elecrooculography is used for tracking eye movements.

ADVANTAGES OF EOG OVER OTHER

Reduces the complexity and cost. Range Linearity Head Movements are Permissible Non-invasive Real-Time

The eye can be modelled as a dipole

ELECTROOCULOGRAPHY

The electrical signal that can be measured from this field is called the Electrooculogram (EOG).
Change in dipole orientation occurs when the eyes move based on which the measured EOG amplitude varies. Analysing these changes eye movement can be tracked. There are two components, i.e. EOGh and EOGv, based on the horizontal and vertical movement of the eye.

EYE MOVEMENT TYPES

SACCADES

The simultaneous movement of both eyes is called a saccade.

Fixations

Fixations are the stationary states of the eyes

BLINKS

Regular opening and closing of the eyelids

ELECTRODE PLACEMENT
Electrodes A & B are used to measure horizontal eye movements Electrodes C & D measure vertical eye movements Electrode E is the ground

EOG CIRCUIT DESIGN

ELECTROOCULOGRAM

ARCHITECTURE FOR EYE BASED ACTIVITY RECOGNITION

EOG SIGNAL PROCESSING

BASELINE DRIFT REMOVAL

Baseline drift is a slow signal change superposing the EOG signal but mostly unrelated to eye movements. Sources- interfering background signals or electrode polarization.
Marginally influences the EOG signal during saccades but influences other eye movements. Wavelet transform-Multilevel 1D decomposition using Daubechies wavelet

Noise Removal
EOG signals

corrupted with noise from different sources, such as the residential power line, the measurement circuitry, electrodes etc.

EOG signals are typically non-repetitive. This prohibits the

application of denoising algorithms that make use of structural and temporal knowledge about the signal
Median filter is used with window size Wmf

Detection of basic eye movements

Saccadic and fixation Detection

Algorithm used is Continuous Wavelet Transform- Saccadic Detection(CWT-SD) Computes continuous 1D wavelet coefficients using mother haar wavelet

Amplitude and direction varies as the activity varies. Different threshold levels are fixed for different activities.

Contd... amplitude SA is the difference in EOG signal amplitude The saccadic

before and after saccade

E.g.: reading involves a fast sequence of small saccades and a large saccade to jump back the beginning of next line. Humans typically alternate between saccades and fixation. So CWT-SD itself can be used for fixation detection. Uses the fact that gaze points remain stable during fixation and they are cluster together closely in time. Can be detected by thresholding on the dispersion of gaze points

Contd...
Blink detection

Algorithm used is continuous wavelet transform-blink detection(CWT-BD) A blink characterised by a sequence of two large peaks in coefficients; one positive and other negative. The time between two peaks is smaller than minimum time between successive saccades rapidly performed in opposite direction. Two saccades have a small fixation in between them.

Analysis for repetitive eye movement patterns

Eye Movement Encoding

It maps the individual saccade information from both EOG components onto a single representation comprised of 24 characters Can be more efficiently processed and analyzed

Wordbook Analysis

Based on the encoded eye movement sequence, it is used to analyse repetitive eye movement patterns.

ELECTRODE PLACEMENT & DIFFERENT OFFICE ACTIVITIES

APPARATUS
Commercial

EOG device- Mobi8 from Twente Medical Systems International(TMSI) Ag/AgCl wet electrodes from Tyco Healthcare placed around the right eye

PERFORMANCE
By

using this technology we get an average precision of 76.1 and an average recall of 70.5.

Human-Computer interaction model

HCI MODEL

APPLICATIONS

Can be used by physically disabled people who have extremely limited peripheral mobility. hands-free operation of static human-computer Assisted Robots Disease recognition.

Interactive gaming systems

LIMITATIONS

only a handful of activities is considered. Precision is found to be only 80%. Additional eye movement characteristics that are potentially useful for activity recognitionsuch as pupil dilation, microsaccades, vestibulo-ocular reflex, or smooth pursuit movementswere not used here because of the difficulty in measuring them with EOG

CONCLUSION

There are two main findings

First, eye movements alone can be used to successfully recognise different activities and can be extended to other activities also other than office activities. Good recognition results were achieved by using feature based algorithm for analysis. More eye movements characteristics should be included

REFERENCES
1) IEEE Transactions on pattern analysis and machine intelligence, vol 33 NO. 4 april 2011 2) http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=5444879

3) P. Turaga, R. Chellappa, V.S. Subrahmanian, and O. Udrea, Machine Recognition of Human Activities: A Survey, IEEE Trans. Circuits and Systems for Video Technology, vol. 18, no. 11, pp. 1473-1488, Nov. 2008.
4) 3. S. Mitra and T. Acharya, Gesture Recognition: A Survey, IEEE Trans. Systems, Man, and Cybernetics, Part C: Applications and Rev., vol. 37, no. 3, pp. 311-324, May 2007.

THANK YOU