logo SBA

ETD

Archivio digitale delle tesi discusse presso l’Università di Pisa

Tesi etd-11262015-094833


Tipo di tesi
Tesi di laurea specialistica
Autore
RAMOS CERECEDA, MILAN DAVID
URN
etd-11262015-094833
Titolo
Design and development of a completely wireless Head Mounted Eye Tracking System on a ARM-based computer running under embedded Linux.
Dipartimento
INGEGNERIA DELL'INFORMAZIONE
Corso di studi
INGEGNERIA BIOMEDICA
Relatori
relatore Scilingo, Enzo Pasquale
relatore Lanatà, Antonio
Parole chiave
  • eye tracking
  • wearable
Data inizio appello
11/12/2015
Consultabilità
Non consultabile
Data di rilascio
11/12/2085
Riassunto
Eye Gaze Trackers (EGTs) systems detect and track the eyes, and their movements, in order to estimate the point-of-gaze. Data obtained from EGTs are utilized in many fields such as ophthalmology, neurology, psychology, marketing, and advertising.
EGTs can be used as diagnostic or assistive instrument. In their diagnostic applications, EGTs can provide objective and quantitative information of the attentional process, emotional state and visual process of a subject. On the other hand, as assistive devices for people with communication disorders or motor disabilities, EGTs can be employed as input for computer interfaces to help people to communicate and interact with the world.
A wearable EGT is a head-mounted device which mounts cameras, light sources or other equipment directly on the user’s head. Wearable EGTs are typically used in light-controlled environments such as laboratories, and require a wired connection between the capture device and the processing module in order to transmit images of the eye.
Commercial EGTs have solved in part those problems but at high prices, around 10000 euros. In contrast, there are EGTs in scientific literature that have reached a high robustness to light changes and a low manufacturing cost. However, studies have not yet dealt with the issue related to wired-connection on EGTs.
The HATCAM is a wearable robust EGT developed in the Biolab Laboratory at IET Department of University of Pisa. The HATCAM has only one camera which is able to capture the user’s eyes and the forward scene simultaneously by means of a mirror. The HATCAM’s camera captures up to 25 frame per second (fps) and has a transmitter which sends images of the eyes up to 30 m. Besides the system lacks of both local control on the images acquisition and reliability on the transmission of images. These technical specifications limit the number of applications of the HATCAM.
The solution can take advantage of current hardware technology which is able to produce embedded devices with high computing performance at low power consumption.
This makes possible to design the capture sub-system on an embedded device and to use of low-cost commercial webcams in order to enhance the capture rate of the HATCAM.
Furthermore, the development of wearable devices can benefit from the Internet of Things (IoT). The IoT is the interconnection of sensors, smart objects and embed- ded devices (“things”) within the existing Internet infrastructure. The IoT allows devices to transfer different and enriched data which can be utilized, for example, for personalized and remote healthcare. The devices of the IoT require low power consumption, and both reliable and real-time wireless communication.
This thesis proposes an improvement to the HATCAM system. The proposed system has two sub-systems: one for acquisition of images of the eyes, and one for processing of those images.
The acquisition sub-system captures images of the eyes and sends them to the processing sub-system. The processing sub-system receives the images and estimates the point-of-gaze in real time. Both sub-systems should be totally wireless to allow the user to wear the system without any obstructive connection cable between the sub-systems. The improved system should employ simple commercial devices as webcam and WiFi adapter in order to have low cost. Moreover, the implemented algorithms have to use all the capabilities of the employed devices to reach high performance.
File