Eye Gaze Tracking Techniques For Human Computer Interaction

The idea of employing the use of eye gaze tracking to evaluate the user experience  has been in existence for quite some time. This stands to be the mind’s guide and the eyes the window of our inner thoughts. The movement of eyes will offer a comprehensive vision into the intentions and thoughts of a person.  The eye movement scrutiny could resolve what one is thinking based on what they are viewing. The evaluation of what one is viewing is done via various techniques in order know what is viewed most. 

Eye Gaze Tracking Techniques For Human Computer Interaction

The video oculography (recording of eye movement and position) techniques receive data/content from an image data of one or more cameras. The detection of the location of the eye in the specified image. Based on the  information acquired from the region of the eye, the gaze direction can be evaluated. 

The tracking and detection of the eye have challenging factors that require certain techniques in order to be effective.  These include the viewing angle, amount of light entering the eye, the eye occlusion, and the positioning of the eye. In video-centric eye tracking, there are 2 different kinds of imaging processes; infrared and visible spectrum imaging. Infrared Eye Tracking basically uses either the dark or white pupil method to evaluate the tracking. 

The Methods

The common eye gaze methods that are applied are either based on the feature or the appearance. 

1. Feature-Centric Evaluation 

The feature-based techniques use the personalities of the human eye to detect a number of features.  These are the most popular features applied in the gaze evaluation. The main point of this technique is to detect the instructive local eye features that are less susceptible to viewpoint and illumination.

The gaze estimation precision will reduce when the precise pupil and iris features aren’t present. The feature-centric techniques also include two more approaches:

• Model-Based

This approach utilizes an open arithmetic eye model in order to evaluate the gaze direction vector in a 3D state. Several geometric approaches count on the metric data and requires a camera calibration and the external light sources, a camera, position and orientation of a monitor. The majority of the model-centric techniques will require to follow one strategy. 

The strategy of eye gaze tracking involves the reconstruction of an optical axis, into 3D, then the reconstruction of the visual axis. The last step is to estimate the gaze point by crisscrossing the scene geometry with the visual axis. The rebuilding of the optical axis is accomplished by estimating the pupil center and the cornea. The gaze point is evaluated by describing the direction of the gaze. Once the gaze direction is defined, it will then be integrated with the object information in the scene. The eye gaze directions are evaluated from the iris to the eyeball center, while being taken as vectors.  

• Interpolation-Centric 

This type of method ignores the mapping from the features of the image to the coordinates of the eye gaze from 2D to 3D. The method assumes that the mapping has a certain parametric or a nonparametric form. The polynomial expressions are among the most used techniques in mapping. Interpolation-based techniques shuns from plainly modeling the physiology and the geometry of the eye of a person, rather, it explains the point that is gazed as a general image features functions.

The data of calibration are employed to evaluate the anonymous mapping function coefficients with the help of a numerical fitting procedure like a multiple linear regression. The neural network-centric eye trackers, which take the place of parametric expressions, ignore a nonparametric form.

This is done in order to apply the mapping from 2D to 3D. In these methods, the tracking of the gaze is accomplished by removing the coordinates of a given facial point and sending them via an instructed neutral network. The directed neutral network should have an outputs that are the coordinates of the point that the user is viewing.

2. Appearance-based Evaluation

This technique identifies and tracks the eye on the basis of the appearance of the photometric. It uses the content of an image to evaluate the eye gaze direction by plotting the image data on the coordinates of the screen. The main methods of this technique are centered on the gray scale unit, the morphable model, and the cross-ratio. The appearance-centric method does not need any calibration of geometry data or cameras. This is because the plotting is accomplished directly on the contents of the image. 

Because there are various methods for tracking the eye gaze, you need to choose the method that is most beneficial. The method used will evaluate the area that the eye visits most on your site, which in turn gives you an idea of what to focus on. Having the estimation and evaluation of the tracking, you can use that to build more similar content in the design of your website. As a person interacts with the computer, special cameras, that are incorporated with an application will perform the tracking.

Such applications will then record the number of times a certain section has been visited. You will also know where the user viewed the most if at all, via the app, you will learn what to invest in and what not to. The sturdiness and precision of data that is captured through eye gaze tracking needs to be enhanced, but is a precise method you should utilize.

bnr14

Megan Wilson
Megan Wilson is user experience specialist & editor of UX Motel. She is also the Quality Assurance and UX Specialist at WalkMe Megan.w(at)walkme.com
Megan Wilson on sabtwitterMegan Wilson on sablinkedinMegan Wilson on sabgoogleMegan Wilson on sabfacebook