警用指纹识别系统中图像预处理的研究

发布时间：2019-01-07 06:46

【摘要】： 指纹鉴别作为一种身份鉴别方法,是人体生物特征鉴别技术中最重要的一种。指纹自动识别技术作为现今生物特征识别技术的主流,有着十分广泛的应用前景。随着近年来计算机技术和网络建设的快速发展,自动指纹识别技术已经越来越多的进入到人们的生活和工作中,自动指纹识别系统的研发和升级也正在成为国内外学术界和商业界的热点。但是,目前的指纹自动识别技术还远远不够成熟,在许多关键步骤的算法上都有着十分广阔的研究空间。在刑事侦查领域,指纹是非常重要的证据,也是查找和确认犯罪人员最直接、最准确的途径。目前,随着指纹自动识别系统在公安机关的推广和普及,已经产生出了巨大的应用效益。但是,在刑侦工作的实践当中,由于指纹来源复杂,而且相当部分的指纹图像质量低劣,给指纹图像预处理增加了难度,直接制约了指纹自动识别系统的性能发挥。在大量关于指纹自动识别技术的研究中,有关此方面的研究报道还不多见。本论文简要分析了警用指纹自动识别系统的基本技术原理,并据此重点针对指纹图像的预处理环节进行了研究。首先,提出了适合于不同来源指纹图像的方向法和方差法相结合的复合分割方法,其次,为提高系统后续处理的可靠性,将基于Gabor滤波的方法应用于指纹图像增强。在此基础上,结合指纹图像的分割面积和方向计算提出了指纹图像的质量评估方法。利用matlab进行的实验表明,本论文提出的图像分割方法能够较好地对各种来源的指纹图像进行分割,克服了大部分指纹图像存在的对比度低,干扰严重等缺陷;基于Gabor滤波的图像增强方法能够显著改善指纹图像的视觉效果;本论文提出的质量评估方法能够正确地排除活体采集过程中的各种低质量的指纹图像。
[Abstract]:Fingerprint identification, as an identification method, is the most important technology of human biometric identification. As the mainstream of biometric identification technology, automatic fingerprint identification technology has a very wide application prospect. With the rapid development of computer technology and network construction in recent years, automatic fingerprint identification technology has been more and more into people's life and work. The research and development of automatic fingerprint identification system (AFIS) is becoming a hot spot in academic and business circles at home and abroad. However, the current automatic fingerprint identification technology is still far from mature, in many key steps of the algorithm has a very broad research space. In the field of criminal investigation, fingerprint is a very important evidence, and also is the most direct and accurate way to find and confirm the criminal. At present, with the popularization and popularization of fingerprint automatic identification system in public security organs, it has produced enormous application benefits. However, in the practice of criminal investigation, because of the complexity of fingerprint source and the inferior quality of fingerprint image, it increases the difficulty of fingerprint image preprocessing and directly restricts the performance of automatic fingerprint identification system. In a large number of researches on automatic fingerprint identification technology, there are few reports on this aspect. In this paper, the basic technical principle of police fingerprint automatic identification system is briefly analyzed, and the preprocessing of fingerprint image is studied. Firstly, a hybrid segmentation method, which is suitable for the combination of directional method and variance method, is proposed. Secondly, in order to improve the reliability of system follow-up processing, the method based on Gabor filter is applied to fingerprint image enhancement. On this basis, a method of fingerprint image quality evaluation is proposed by combining the segmentation area and direction calculation of fingerprint image. The experiments conducted by matlab show that the image segmentation method proposed in this paper can segment fingerprint images from various sources well, and overcome the shortcomings of low contrast and serious interference in most fingerprint images. The image enhancement method based on Gabor filter can significantly improve the visual effect of fingerprint image, and the quality evaluation method proposed in this paper can correctly eliminate all kinds of low-quality fingerprint images in the process of living body acquisition.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2007
【分类号】：D918.91

【引证文献】