基于智能手机终端的指纹识别算法优化实现

发布时间：2018-06-07 16:10

本文选题：智能手机 + 信息安全　；参考：《吉林大学》2015年硕士论文

【摘要】：随着移动互联网技术和智能手机终端的不断完善，智能手机移动支付受到了前所未有的关注。而随着通信和信息化水平的提高，人们也越来越普遍地使用手机来存储和传输大量的个人信息与隐私，并且希望保证个人信息安全。指纹识别技术是目前应用于智能手机终端的最为成熟的身份认证与支付交易认证技术，研究基于智能手机终端的指纹识别算法具有十分广泛的重要意义。指纹识别效果的好坏，除了取决于指纹采集技术的高低，更在很大程度上取决于指纹识别算法的优劣。预处理过程是指纹识别中重要的一环。所以在本文中，作者将着重针对预处理过程中涉及到的三个主要步骤进行研究：图像增强、二值化和细化。我们会采用FVC2004（Fingerprint Verification Competition2004,2004指纹识别竞赛）的DB2和DB4指纹数据库来对指纹进行性能测试。经FVC2004指纹性能测试的仿真试验验证，我们可以得出如下结论：基于相同的特征提取和匹配算法，本文提出的指纹图像预处理优化算法更加高效，能比传统算法更好地消除噪音干扰，避免各种伪特征信息。针对上述问题本文拟开展如下研究： 1.基于傅里叶和方向滤波的指纹图像增强算法基于Gabor滤波的增强算法，对较高质量指纹图像处理效果好，但占用预处理过程的时间长。基于傅里叶滤波的增强算法，速度快并对低质量指纹图像的增强效果好。本文采用基于傅里叶变换和方向滤波的指纹图像增强算法，实验证明能够取得更好的指纹图像处理效果。 2.基于方向图场的动态阈值二值化算法基于固定阈值的二值化算法，虽然算法简单，但处理质量差的指纹图像会产生伪特征信息；基于动态阈值的二值化算法，虽然在某些方面能够弥补固定阈值算法的不足，，但也会引入噪音。本文采用基于图场的动态阈值算法，经实验证明可以取得良好的抗噪性。 3.基于改进的OPTA细化算法快速细化算法利用像素的4领域判断指纹边界点并逐步删除；OPTA细化算法利用8个消除模板和2个保留模板进行对比，判断中心像素是否删除；改进的OPTA细化算法改造了算法的消除模板和保留模板。本文将在进行改进的OPTA细化算法的基础上，对细化后的指纹图像采用方形模板进行纹线跟踪，实验证明与传统算法相比较，本文提出的算法能够更好地去除伪特征点、克服干扰因素。
[Abstract]:With the continuous improvement of mobile Internet technology and smart phone terminals, mobile payment has received unprecedented attention. With the improvement of communication and information technology, mobile phones are more and more widely used to store and transmit a large amount of personal information and privacy, and they hope to ensure the security of personal information. Fingerprint identification technology is the most mature authentication and payment authentication technology used in smart phone terminal. It is very important to study fingerprint identification algorithm based on smart phone terminal. The effect of fingerprint recognition depends not only on the level of fingerprint acquisition technology, but also on the advantage and disadvantage of fingerprint recognition algorithm to a great extent. Pretreatment is an important part of fingerprint recognition. Therefore, in this paper, the author will focus on three main steps involved in the process of preprocessing: image enhancement, binarization and thinning. We will use the DB2 and DB4 fingerprint databases of FVC2004(Fingerprint Verification Competition 2004 to test fingerprint performance. Through the simulation of FVC2004 fingerprint performance test, we can draw the following conclusions: based on the same feature extraction and matching algorithm, the fingerprint image preprocessing optimization algorithm proposed in this paper is more efficient. It can eliminate noise interference and avoid all kinds of pseudo-feature information better than traditional algorithm. In view of the above problems, this paper intends to carry out the following research: 1. Fingerprint Image Enhancement algorithm based on Fourier and Direction filtering The enhancement algorithm based on Gabor filter has good effect on fingerprint image processing with high quality, but it takes up a long time in the process of preprocessing. The enhancement algorithm based on Fourier filter has high speed and good enhancement effect on low quality fingerprint image. In this paper, a fingerprint image enhancement algorithm based on Fourier transform and directional filtering is proposed, and the experimental results show that it can achieve better fingerprint image processing results. 2. Dynamic threshold binarization algorithm based on pattern Field Although the binarization algorithm based on fixed threshold is simple, it can produce pseudo-feature information when dealing with bad quality fingerprint image, while the binarization algorithm based on dynamic threshold can make up for the deficiency of fixed threshold algorithm in some aspects. But it also introduces noise. The dynamic threshold algorithm based on graph field is used in this paper. 3. Improved OPTA thinning algorithm The fast thinning algorithm uses the four fields of pixels to judge the fingerprint boundary points and gradually deletes the OPTA thinning algorithm compared with 8 elimination templates and 2 reserved templates to judge whether the center pixels are deleted or not. The improved OPTA thinning algorithm has modified the elimination template and reserved template of the algorithm. Based on the improved OPTA thinning algorithm, this paper uses the square template to track the lines of the thinned fingerprint image. The experimental results show that the algorithm proposed in this paper can remove the pseudo-feature points better than the traditional algorithm. Overcome interference factors.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.41;TN929.53

【参考文献】