信息安全研究 ›› 2016, Vol. 2 ›› Issue (1): 40-43.

• 生物特征识别专题 • 上一篇    下一篇

虹膜识别技术进展与趋势

李海青   

  1. 中国科学院自动化研究所
  • 收稿日期:2015-12-18 出版日期:2016-01-05 发布日期:2016-01-18
  • 通讯作者: 李海青
  • 作者简介:李海青 助理研究员,主要研究方向为生物特征识别、计算机视觉和机器学习. hqli@nlpr.ia.ac.cn 孙哲南 研究员,主要研究方向为生物特征识别、人工智能和机器学习. znsun@nlpr.ia.ac.cn 谭铁牛 研究员,主要研究方向为生物特征识别、图像与视频理解和信息取证与安全. tnt@nlpr.ia.ac.cn 何召锋 副教授,主要研究方向为生物特征识别、计算机视觉和机器学习. hezhf@irisking.com 马力 教授级高级工程师,主要研究方向为生物特征识别、语义网和人工智能. mali@irisking.com

Progress and Trends in Iris Recognition

Li Haiqing   

  • Received:2015-12-18 Online:2016-01-05 Published:2016-01-18

摘要: 虹膜识别具有很高的精度和稳定性,已广泛应用在金融、边防和门禁等领域.经过20多年的发展,虹膜识别在成像装置和识别算法方面均取得了显著的进展.一方面,虹膜成像装置的成像距离越来越远、成像范围越来越大、重量体积越来越小,明显提高了虹膜识别系统的易用性.另一方面,大规模的应用促进了许多低质量虹膜图像处理、快速分类检索、跨设备识别和安全隐私保护方法的研究.未来几年,虹膜识别将在技术、应用和行业等方面呈现出以下六大发展趋势:从近红外到多光谱、从人工设计到数据驱动、从人配合机器到机器配合人、从固定设备到移动互联、从可控环境到复杂场景、从各行其是到标准规范.

关键词: 虹膜识别, 虹膜成像装置, 虹膜分割, 质量评价, 特征分析, 跨设备识别

Abstract: Iris recognition is very accurate and stable and has been widely applied in finance, border control and access control, etc. Much progress has been achieved both in iris imaging and recognition algorithms in the past two decades. On one hand, iris imaging devices are with further standoff distance, larger capture volume and smaller size, which greatly improves the ease of use of iris recognition systems. On the other hand, large scale applications promote the research in low quality iris image processing, fast classification and retrieval, crosssensor recognition, data security and privacy protection. In the next several years, iris recognition will present the following six trends in technology, application and industry: from near infrared to multispectral illumination, from rule based to data driven algorithm development, from passive to active user interaction, from fixed and standalone devices to mobile and connected devices, from constrained to complex applications, from disordered to standardized industry.

Key words: iris recognition, iris imaging devices, iris segmentation, quality assessment, feature analysis, crosssensor recognition