Please use this identifier to cite or link to this item:
Title: Optimization of Error in Localization of Iris for Non-Frontal Faces
Authors: Kumar, Ipsita
Supervisor: Kumar, Sukhwinder
Keywords: eye localization;eye tracking;skin segmentation;template matching
Issue Date: 12-Sep-2019
Abstract: Eye Localization and eye tracking has been a vast area of research since a while now. There are a lot of applications that works on the basis of eye tracking and eye localization such as in the case of medical identification, human computer interface, gaming and entertainment and so on. The field of eye tracking consist of an issue till date that is eye tracking at non frontal faces or eye localization at different face positions. In this thesis research we try to improve the accuracy of eye localization at different face positions thus, resolving this issue. The method carried out in this research is on the basis of coordinate projections followed by template matching and skin segmentation. The problem also comes in the case of detection of face specially where it is non frontal. In order to carry out the process of eye localization in a smooth manner, we tend to extract the face using the process of skin segmentation. Further, with the help of template matching eye area is extracted, which leads to the localization of eyeball. The process assures the extraction of face at different positions, making it an advantage of skin segmentation. Later we present the results of each part of the proposed methodology one by one. In the end the results are optimized, the accuracy and losses are determined to show how effectively the method worked. This can enhance a lot in the area of research such as determining the eye in the case of occlusions and reflections at non frontal faces. Another upgradation that can be made in this field is determination of eye ball in the shaking situation. A lot of applications can be further upgraded and a lot of new applications can be accomplished with the help of eye ball localization in non-frontal faces.
Appears in Collections:Masters Theses@ECED

Files in This Item:
File Description SizeFormat 
Ipsita Final Report 801761007.pdf2.11 MBAdobe PDFThumbnail

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.