แจ้งเอกสารไม่ครบถ้วน, ไม่ตรงกับชื่อเรื่อง หรือมีข้อผิดพลาดเกี่ยวกับเอกสาร ติดต่อที่นี่ ==>
หากไม่มีอีเมลผู้รับให้กรอก thailis-noc@uni.net.th
หากไม่มีอีเมลผู้รับให้กรอก thailis-noc@uni.net.th
Suwitchaya Rattarom. Performance improvement of gaze mapping algorithm in low-cost gaze tracking system. Doctoral Degree(Computer Engineering). Mae Fah Luang University. The Learning Resources and Education Media Center. : , 2561.
Performance improvement of gaze mapping algorithm in low-cost gaze tracking system
Name: Suwitchaya Rattarom
Address:
333 Moo1, Thasud, Muang, Chiang Rai 57100
Organization :
Mae Fah Luang University. School of Information Technology. Computer Engineering
Email :
ask.library@mfu.ac.th
keyword:
Tracking system
Classification :.LCCS: QA76.9.H85
LCSH:
Eye tracking
Abstract:
Gaze tracking system, the system that uses human gazing as the input for computer
interaction, is one of the most important topics for many researches, for instance, human
computer interaction, user interface design, advertising, save driving, and neuroscience.
In the past decade, many researchers have been trying to make the system suitable
for most people by using a low-cost camera like web camera instead of a high performance
camera. However, they need more study to make the system acceptable because low-cost
hardware produces less accuracy than the hi-end hardware.
This dissertation, hence, proposes the methods for performance improvement of
low-cost gaze tracking system by using three salient features of the eyes. The features
consist of a center of pupil, glint reflection, and inner eye corner. The methods are
divided into two scenarios: head fixing and natural head pose. For the situation of the
head it remains stationary, on a feature based gaze estimation method. We propose
the construction and validation of the polynomial models for low-cost gaze tracking
system according to the statistical data analysis. The Proposed models are constructed by
(4)
seeking for the relationship between the position of the calibration targets and position
of pupil-glint vector. After that, the statistical hypothesis testing is used to validate the
model. The state-of-the-art model is used for comparing with our proposed ones. The
results reveal that the proposed model obtains a higher accuracy than the existing models.
For the situation of natural head pose, we propose a novel training vector which is used
to classify the regions of interest. The vector is derived from combining 3 features of
the eyes into 22-features-training vector. Number of machine learning techniques were
applied for testing the performance of the obtained training vectors. The results show that
the artificial neural network with 2 hidden layers gives the best classification performance.
The accuracy is acceptable in many application, especially when compared to the cost of
our system and simplicity of imitation.
Mae Fah Luang University. The Learning Resources and Education Media Center
Address:
Chiang Rai
Email:
library@mfu.ac.th
Name: Nattapol Aunsri
Role:
Advisor
Email :
ask.library@mfu.ac.th
Name: Surapong Uttama
Role:
Co-Advisor
Email :
ask.library@mfu.ac.th
Created:
2561
Modified:
2019-05-21
Issued:
2562
Issued:
2019-05-21
วิทยานิพนธ์/Thesis
application/pdf
CallNumber:
Thesis QA76.9.H85 S967p 2018
eng
DegreeName: Dotor of Philosophy
Level: Doctoral Degree
Descipline: Computer Engineering
©copyrights Mae Fah Luang University
RightsAccess:
ใช้เวลา
0.038738 วินาที
Suwitchaya Rattarom
| Title | Contributor | Type |
|---|---|---|
| Performance improvement of gaze mapping algorithm in low-cost gaze tracking system
มหาวิทยาลัยแม่ฟ้าหลวง Suwitchaya Rattarom | Nattapol Aunsri Surapong Uttama | วิทยานิพนธ์/Thesis |
Nattapol Aunsri
| Title | Creator | Type and Date Create |
|---|---|---|
| Adaptive genetic algorithms for particle filtering improvement
มหาวิทยาลัยแม่ฟ้าหลวง Nattapol Aunsri | Chanin Kuptametee | วิทยานิพนธ์/Thesis |
| Performance improvement of gaze mapping algorithm in low-cost gaze tracking system
มหาวิทยาลัยแม่ฟ้าหลวง Nattapol Aunsri;Surapong Uttama | Suwitchaya Rattarom | วิทยานิพนธ์/Thesis |
| Integrated action crossing method for drug-drug interactions prediction in non-communicable diseases
มหาวิทยาลัยแม่ฟ้าหลวง Nattapol Aunsri;Thongchai Yooyativong | Sathien Hunta | วิทยานิพนธ์/Thesis |
| Discrete event system simulation for Intra-hospital patient transfer improvement
มหาวิทยาลัยแม่ฟ้าหลวง Nattapol Aunsri | Ekkarat Meephu | วิทยานิพนธ์/Thesis |
| Human face skin type classification using clahe technique and deep learning framework
มหาวิทยาลัยแม่ฟ้าหลวง Nattapol Aunsri | Sirawit Saiwaeo | วิทยานิพนธ์/Thesis |
Surapong Uttama