计算机工程

• 图形图像处理 • 上一篇    下一篇

基于变分水平集模型的虹膜图像分割方法

张荷萍,徐效文   

  1. (中南大学地球科学与信息物理学院生物医学工程研究所,长沙 410083)
  • 收稿日期:2012-04-05 出版日期:2013-10-15 发布日期:2013-10-14
  • 作者简介:张荷萍(1986-),女,硕士研究生,主研方向:模式识别,生物特征识别;徐效文(通讯作者),副教授、博士
  • 基金项目:
    中央高校基本科研业务费专项基金资助项目(201012200064)

Iris Image Segmentation Method Based on Variational Level Set Model

ZHANG He-ping, XU Xiao-wen   

  1. (Institute of Biomedical Engineering, School of Geosciences and Info-physics, Central South University, Changsha 410083, China)
  • Received:2012-04-05 Online:2013-10-15 Published:2013-10-14

摘要: 当虹膜图像受眼睑、睫毛及变形等影响时,会造成分割不准确,如何有效地分割虹膜是虹膜识别面临的主要问题。为此,提出一种基于变分水平集模型的虹膜图像分割方法。采用灰度投影法粗略确定瞳孔,使用最小二乘拟合得到瞳孔与虹膜边界,通过变分水平集模型精确分割虹膜。实验结果表明,该方法对虹膜外边缘分割的准确率为98.59%,高于Daugman、Hough变换及改进Hough变换等常用方法。

关键词: 虹膜分割, 变分水平集, 最小二乘拟合, 能量函数, 水平集演化方程, 变形虹膜

Abstract: Segmenting the iris accurately is a main problem for iris recognition, due to the impact of the eyelids, eyelashes and deformation. This paper presents an iris segmentation method based on variational level set model. It uses gray projection algorithm to locate the pupil, applies a least square fitting algorithm to estimate the boundary between the pupil and the iris, and employs a variational level set model to accurately segment the iris. Experimental results demonstrate the segmentation accuracy of 98.59% for outer edge of the iris is better than using Daugman method, Hough transformation method and improved Hough transformation method.

Key words: iris segmentation, variational level set, least square fitting, energy function, level set evolution equation, deformed iris

中图分类号: