Academic Journal

Evaluating the impact of image preprocessing on iris segmentation

التفاصيل البيبلوغرافية
العنوان: Evaluating the impact of image preprocessing on iris segmentation
المؤلفون: José F. Valencia-Murillo, Daniel A. Poveda-Sendales, Daniel F. Valencia-Vargas
المصدر: TecnoLógicas, Vol 17, Iss 33, Pp 31-41 (2014)
بيانات النشر: Instituto Tecnológico Metropolitano, 2014.
سنة النشر: 2014
المجموعة: LCC:Technology
LCC:Engineering (General). Civil engineering (General)
مصطلحات موضوعية: Algoritmos, biometría, imágenes del iris, preprocesamiento de imágenes, segmentación., Technology, Engineering (General). Civil engineering (General), TA1-2040
الوصف: Segmentation is one of the most important stages in iris recognition systems. In this paper, image preprocessing algorithms are applied in order to evaluate their impact on successful iris segmentation. The preprocessing algorithms are based on histogram adjustment, Gaussian filters and suppression of specular reflections in human eye images. The segmentation method introduced by Masek is applied on 199 images acquired under unconstrained conditions, belonging to the CASIA-irisV3 database, before and after applying the preprocessing algorithms. Then, the impact of image preprocessing algorithms on the percentage of successful iris segmentation is evaluated by means of a visual inspection of images in order to determine if circumferences of iris and pupil were detected correctly. An increase from 59% to 73% in percentage of successful iris segmentation is obtained with an algorithm that combine elimination of specular reflections, followed by the implementation of a Gaussian filter having a 5x5 kernel. The results highlight the importance of a preprocessing stage as a previous step in order to improve the performance during the edge detection and iris segmentation processes.
نوع الوثيقة: article
وصف الملف: electronic resource
اللغة: English
Spanish; Castilian
تدمد: 0123-7799
2256-5337
Relation: http://itmojs.itm.edu.co/index.php/tecnologicas/article/view/599; https://doaj.org/toc/0123-7799; https://doaj.org/toc/2256-5337
URL الوصول: https://doaj.org/article/1db6710720aa4d3d87c866e3d44999b9
رقم الانضمام: edsdoj.1db6710720aa4d3d87c866e3d44999b9
قاعدة البيانات: Directory of Open Access Journals