期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2016
卷号：94
期号：2
出版社：Journal of Theoretical and Applied
摘要：Iris pattern is one of the most consistent biometric methods used for recognizing and identifying persons. Employing videos as a capturing instrument is a pretty modern style in the area of iris biometric. The use of frame by frame method provides more information and offers more suppleness compared to old-fashioned still images. Nevertheless, the size, quality and shape of the iris might differ between a frame and another. Additionally, to getting best performance it need a rapid and precise method to segment iris to amelioration rate of recognition. This work presents a method for choosing the best frames found in an iris video. This method is based on detecting motion blur and occlusion in iris videos and investigating their influence on the process of recognition. This proposed is followed by a rapid and precise method to detect pupil area, this method on the grounds of dynamic threshold with Circular Hough Transform then apply Geodesic Active Contour for detect outer boundary of iris. Experimental results are carried out on the MBGC NIR Iris Video datasets from the National Institute for Standards and Technology (NIST). Results show that the suggested selection method in NIR Iris Videos results in substantial enhancement in recognition efficiency. Results also indicated that the experimental evaluation of Iris segmented technique proposed in this work indicates that the precision and speed of the iris recognition via video is improved.
