【作者】 杨涛；

【导师】 吕林涛；

【作者基本信息】 西安理工大学 ， 计算机应用技术， 2009， 硕士

【摘要】 随着互联网技术的高速发展,身份认证技术得到了广泛的应用,传统的密码认证技术已经不满足当今社会的要求,生物特征识别技术应运而生。虹膜识别技术以其独特的生物识别优越性成为信息安全领域的研究热点,虹膜识别的研究涉及到计算机视觉、数字图像处理、模式识别等众多学科领域。本文针对虹膜识别中的特征提取技术及相关算法进行重点研究。通过对传统滤波器算法的分析研究,由于虹膜纹理的固有复杂度及滤波器的构造缺陷,会导致滤波器的构造不能被普遍适用。针对滤波器存在的问题,本文在2D Log-Gabor滤波器的基础上提出了一种二维奇对称Log-Gabor滤波器改进算法。首先将2DLog-Gabor滤波器进行坐标系转换,屏蔽了角度方向带宽过大的影响,其次对滤波器进行奇对称变换,提高有效特征提取率,最后对提取出的系数矩阵进行特征融合,从而减少冗余特征数量。该算法通过对虹膜纹理的相位和幅度信息在不同频率和方向进行分析,采用结构化纹理分析找出空间域中虹膜纹理的灰度分布规律,采用特征融合技术减少海明距离总编码长度,提高有效编码率,有效地解决了传统滤波器存在的问题。通过实验对改进算法进行验证,并设计实现了虹膜识别系统。实验结果表明,本算法对噪声和干扰具有较强的鲁棒性,能有效提高虹膜识别率,具有较高的理论和实用价值。更多还原

【Abstract】 With the rapid development of Internet technology, Identity Authentication technology has been widely used, the traditional password authentication technology have can’t been adapted the requirement of security today. Biometric identification technology came into being. Iris identity authentication technology has been wildly researched in security fields by its unique advantages. Iris recognition research related to multidisciplinary, such as computer vision, digital image processing, pattern recognition and many other subject fields. So iris recognition have great mean in researching. Feature extraction is mainly researched in this paper.Researched by the traditional wavelet filter algorithms, because of the inherent complexity of iris texture and the filter construct defects, not all filters can be used in every condition. Aiming at the defects of traditional filters, an improved algorithm which is odd symmetry 2D Log-Gabor filter is proposed in this article, the algorithm is based on 2D Log-Gabor filter. Firstly, transform the coordinate of original filter; it shielded the influence that the angle direction of bandwidth is always overflowing. Secondly, extract odd symmetry filter from the transformed filter, the effective rate of feature extraction are raised by the process. Finally, in order to reduce the number of redundant features, the extracted feature coefficient matrix is fused. The algorithm analyzed the phase position and amplitude of iris texture in different frequency and directions; the iris spatial grayscale distribution can be found by structured texture analysis, and take feature fusion to reduce the length of Hamming distance, enhance coding rate and solve the problem of traditional filters effectively.By taking experiments, the improving algorithm is verified; and iris recognition system is designed and finished. The experimental result shows that the odd symmetry 2D Log-Gabor filters have strong robustness with noise and interferences. This algorithm has a high value of theoretical and practical.更多还原