【作者】 薛亚许；

【导师】 陈金玉；

【作者基本信息】 重庆大学 ， 控制科学与工程， 2011， 硕士

【摘要】 人类应用指纹的历史开始于远古。指纹是最古老的身份证,早在6000年前人类就会利用指纹来代替签名。考古学家证实,早在原始社会晚期,指纹作为身份鉴别的工具已经在我国开始应用。尽管指纹在我国古代应用广泛,但这种应用仅是一种感性应用,没有在此基础上进行分析综合和归纳推理,更没上升到科学及理论应用。随着社会发展和科学技术的进步,人们对指纹模式识别系统的研究越来越深入,并取得了巨大的进展。比如在很多企业和学校引入了指纹考勤系统,但是目前的考勤系统在识别速度和准确性上仍然存在一些问题,因此就要求我们对指纹模式识别算法做更进一步的研究。本文在通过对国内外有关指纹识别技术研究成果的学习基础上,对指纹模式识别系统算法进行了研究。主要包括:指纹图像预处理、指纹图像特征点提取和指纹模式特征匹配。论文详细地分析了国内外指纹模式识别技术的发展状况,在总结和学习前人的理论基础上,以提高指纹模式识别算法处理速度和性能作为主要研究目标,重点对指纹二值化算法及指纹模式特征匹配算法进行了深入的研究。首先,针对目前二值化算法存在的抗燥能力差、阈值不易选取等问题以及效率低的不足,提出了一种新的二值化灰度阈值分割法,该算法很好地体现了易实现和计算量小的特点,且把局部自适应和指纹灰度图像信息相结合,同时也起到了一定的抗干扰能力。接着针对指纹模式特征匹配,提出了一种新的基于遗传算法的二次匹配算法,其中的细节特征点初筛部分,提出了一种新的构建与变换无关的结构信息的方法;坐标校准部分,采用极坐标法;对于校准后得到的特征点,采用可变大小的限界盒方法,提高了算法效率。最后采用FVC2004指纹库中的指纹图像,通过与经典指纹模式特征匹配算法的结果比较分析,验证了本文算法的可行性。更多还原

【Abstract】 Application of the fingerprint of human history began in the ancient. Fingerprint identification is the oldest, as early as 6000 years ago, will use fingerprints to replace the human signature. Archaeologists confirmed that, as early as the late primitive society, the fingerprint as identification tool has been applied in China.While fingerprints are widely used in ancient China, but this application is only an emotional application, not on the basis of this synthesis and inductive reasoning and did not rise to the science and application of theory. With the social development and scientific and technological progress, the people of fingerprint pattern recognition system, more and more in-depth research, and have made tremendous progress. Many enterprises and schools such as the introduction of the fingerprint attendance system, but the current time and attendance system in recognition speed and accuracy are still some problems, and therefore requires us to fingerprint pattern recognition algorithms to do further study.In this paper, research on fingerprint recognition technology in learning rough home and abroad, is studied based on the fingerprint pattern recognition system algorithm. Mainly contains: fingerprint image preprocessing、fingerprint feature extraction and fingerprint matching.In the thesis the development of fingerprint pattern recognition at home and abroad was summarized in detail, fingerprint binarization algorithm and fingerprint pattern character matching algorithm were further investigated based on previous theory in order to improve the processing speed and performance of fingerprint pattern recognition algorithm. Firstly, as binary algorithm had poor anti-noise properties and difficulties in selecting suitable threshold value, a new binarized gray threshold segmentation method, which combined local adaptive and fingerprints grayscale image information and had and the characteristics of less calculation and preferable antijamming, was come up. Then based on genetic algorithm a new re-matching algorithm, of which the detail screening part was figured out by a new mothod of building irrelevant transform structure information and the coordinates calibration part was settled by polar coordinates, was put forward. Then an adjustable Bounding Box is used to process the calibrated feature points. Finally, it was proved by comparing the finger images in the FVC2004 fingerprints database with the images obtained by classic finger print pattern matching algorithm that this algorithm was feasible.更多还原