虹膜特征稳定性提取的关键技术研究

发布时间：2018-06-20 05:26

本文选题：虹膜识别 + 特征稳定性　；参考：《河南科技大学》2017年硕士论文

【摘要】：随着我国信息化技术的深入发展,信息安全日益成为社会关注的重要问题。生物特征识别技术由于自身具有的重要特性,已被广泛的关注和应用,其中虹膜识别技术由于自身的特点(高度准确性、唯一性、稳定性、防伪性和非接触性)而被认为是一种具有高度研究价值和应用前景的生物识别技术。然而在实际的应用中,虹膜识别技术还存在着很多的缺陷,如低质量的虹膜图像干扰、虹膜定位不精确和虹膜特征选取不当等因素都会导致提取的虹膜特征稳定性较差,从而降低整个识别系统的性能。所以对影响虹膜特征稳定性的关键技术研究很有必要。本文首先梳理和分析了国内外虹膜识别技术的研究现状,对当前虹膜识别中的几个主要的虹膜识别算法进行了分析研究,并将这些相关算法进行了比较;然后从整体上详细的介绍了虹膜识别系统的各个部分的内容,以及采用的算法技术;最后详细的介绍了本文在虹膜识别的关键技术上的创新和改进,并在中科院CASIA_V1.0数据库和实验室自建数据库BA上对改进的算法进行了实验分析。本文的主要研究内容如下:1.为了消除低质量的虹膜图像(采集设备和采集环境的差异)对识别系统性能的影响,本文采用了基于Fourier变换的虹膜图像质量评价算法,去除采集的低质量图像,降低虹膜识别系统的误识率。2.针对虹膜外圆定位时受图像质量影响较大,搜索范围增大耗时较长的问题,研究了一种结合统计知识和微分积分定位的虹膜外圆定位算法,缩小了圆周的定位精度,提高了定位速度。3.仔细的研究了虹膜纹理的局部细节特征,提出了一种利用多通道的二维Gabor滤波器对虹膜子块局部细节进行量化的特征提取方法,并结合相位信息组合成特征向量,以此来编码虹膜特征。对数据库中的图像样本进行实验,实验结果表明,本文采用的基于Fourier变换的虹膜图像质量评价算法、改进的虹膜外圆定位算法和基于关键点间特征向量的虹膜特征提取算法,使得提取的虹膜特征具有较高的稳定性,并提高了识别系统的速率和识别正确率,符合现有虹膜识别系统开发技术要求。
[Abstract]:With the further development of information technology in China, information security has become an important issue of social concern. Biometric recognition technology has been widely paid attention to and applied because of its own important characteristics, among which iris recognition technology is highly accurate, unique and stable. It is considered to be a biometric technology with high research value and application prospect. However, in practical application, there are still many defects in iris recognition technology, such as low quality iris image interference, inaccurate iris location and improper selection of iris features, which will lead to poor stability of extracted iris features. Thus, the performance of the whole recognition system is reduced. Therefore, it is necessary to study the key techniques that affect the stability of iris features. In this paper, firstly, the research status of iris recognition technology at home and abroad is analyzed and analyzed, and several main iris recognition algorithms are analyzed and compared. Then the content of each part of iris recognition system is introduced in detail, and the arithmetic technology is introduced in detail. Finally, the innovation and improvement of the key technology of iris recognition in this paper are introduced in detail. The improved algorithm is analyzed experimentally on CASIA V1.0 database and lab database BA. The main contents of this paper are as follows: 1. In order to eliminate the influence of the low quality iris image (the difference between the acquisition equipment and the acquisition environment) on the performance of the recognition system, an iris image quality evaluation algorithm based on Fourier transform is adopted in this paper to remove the collected low quality image. Reduce the error rate of iris recognition system. 2. Aiming at the problem that the iris outer circle location is greatly affected by image quality and the search scope is increased, a new iris circle location algorithm combining statistical knowledge and differential integral positioning is studied, which reduces the accuracy of the circle location. Improved positioning speed. 3. In this paper, the local details of iris texture are studied carefully, and a feature extraction method based on multi-channel two-dimensional Gabor filter is proposed to quantize the local details of iris sub-block, and the feature vector is formed by combining the phase information. This is used to encode iris features. The experimental results on the image samples in the database show that the iris image quality evaluation algorithm based on Fourier transform, the improved iris outer circle location algorithm and the iris feature extraction algorithm based on the feature vectors between key points are adopted in this paper. The extracted iris features are of high stability, and the rate and accuracy of the recognition system are improved, which meets the technical requirements of the existing iris recognition system.
【学位授予单位】：河南科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP309

【参考文献】