【作者】 彭智勇；

【导师】 曾庆宁；

【作者基本信息】 桂林电子科技大学 ， 信号与信息处理， 2006， 硕士

【摘要】 随着社会经济的发展,目前基于虹膜的身份识别技术是国内外发展很快的一研究热点,并显示了很大的优越性:虹膜是独一无二的个人特征;虹膜特征在人一生中保持高的稳定性;能够对虹膜进行准确的活体检测,有效地防止人工伪造;虹膜图像的采集具有无侵犯性等等。本文详细介绍了已有的两种较成熟的虹膜识别系统:Daugman的虹膜识别系统与Wildes的虹膜识别系统。较深入地研究了虹膜识别算法,并实现了一虹膜考勤系统。本文主要讨论了对虹膜采集装置的改进、定位、图像质量评定、活体检测和编码以及测试统计。在虹膜采集装置方面,提出了利用波阵面扩展采集装置景深的方法。使用波阵面能有效降低采集装置对景深的敏感性,使在离焦距±15cm的范围里采得的图片都能用于身份识别。在定位方面,改进了Daugman的虹膜定位算法,使定位的速度和定位出的虹膜的有效性都有所提高。对瞳孔与巩膜的定位都是分两步进行的,并且对定位的准确性能进行有效检测。对上下眼皮的定位,采用直接去除可能的眼皮区域的方法。在虹膜质量评定方面,作者提出了新的虹膜质量评定方法。通过对图像总体质量的评价、虹膜是否处于图片中心的检测、定位后的虹膜部分质量的检测,从而对虹膜图像质量进行了有效的评定。在虹膜的活体检测方面,提出了利用瞳孔放缩、角膜亮度或能量谱进行活体检测的方法。在编码方面,改进了Daugman的Gabor小波编码算法:对每一象素的特征码加入了有效性标志位,相应地对匹配方法也进行了改进,即只对匹配双方都有效的象素点进行匹配,从而降低了拒识率。在测试统计方面,对本文提出的算法进行了大量测试,主要是对错误接受率与错误拒绝率进行了有效统计,统计结果证明了算法的有效性。最后指出了当今虹膜识别系统还需解决的主要问题,为今后的研究提出了作者的建议。更多还原

【Abstract】 Iris recognition system has been an active research area in recent years as a result of the social and economical development, and it shows great advantages. For example, iris posses distinct features for uniquely identifying a person, the characteristics of one’s iris are steady all one’s life, can check whether is nature iris exactly, which prevents counterfeiting. The capturing of iris image is non-invasive, and so on.In this paper, a detail discussion of the two mature iris recognition systems, Daugman’s and Wildes’s, is carried out. We have done deeply research on iris recognition algorithm and implemented a detection system on work attendance. The researches are mainly focused on the improvement of iris sampling equipment, iris localization, the quality evaluation of iris image, live detection, coding and test statistic.A method using wavefront to extend the depth-of-field of the image capture system is expatiated in the paper. To use wavefront can reduce the sampling system’s sensitivity to depth-of-field, which make the image sampled in the range of±15cm available for ID recognition.The iris localization is the key step for iris recognition, an improvement localization algorithm based on Daugman’s algorithm is proposed, which is provided not only more efficient on localization speed, but also more effective on the performance than the famous Daugman’s algorithm. To localize both upper and lower eyelid boundaries, the method used in this paper excludes directly the area of iris covered by the eyelid.For the quality evaluation of iris image, a new evaluation method is introduced. First, It detects if the iris is in the middle of the image, then evaluate the quality of localized iris part. This method has a effective quality evaluation of the iris image.With regard to the iris live detection, this paper proposed a method. It is done according to the pupil zoom, cornea and energy spectrum.Some improvements are made on Daugman’s Gabor wavelet coding algorithm, we add an effectively flag in the pattern codes of every pixel, so the matching method is modified accordingly, To do the match operations only on the both effective pixels, which reduces the percentage of rejection recognition.A great deal of tests have been done on our algorithm, we statistic the percentage of wrong acceptance and wrong recognition, The test results are also presented and analyzed, which show the validity of our algorithms.In the end of this paper, the main problems to be solved for the current iris recognition system and the future research work to be done in this field are proposed.更多还原