| With the development of information science and technology, the recognition and identification of persons by use of physiological characteristics have become one of the research focuses in computer science and pattern recognition in many countries all over the world. Fingerprint, one of the most stable human physiological characteristics, has been widely used for personal identity recognition in developed countries. In recent years, researches on fingerprint recognition techniques and their applications also raised upsurges in our country. Unfortunately, most of them are focused rather on the secondary development of the techniques purchased from abroad than on the study of self-owned algorithms and their hardware realization. Considering this fact, we will concentrate ourselves on the development of self-owned algorithms and their hardware realization. This thesis aims at the development and presentation of such a fingerprint system. From the research point of view, the difference between them is not clear. But, some developed countries have done a farther research because of an early start, advanced, research methods and equipments. Automatic recognition system has already been widely used in many fields in these countries. Comparatively speaking, our research in this area started relatively late, from around 80S. It emphasized the research angle, so that it was hot used in practice, So,there is long distance to make these techniques meet the market's need. Recently, the research of automatic fingerprint recognition concentrates on preprocessing, characteristics acquisition, classification of fingerprint images, researches in correspondent algorithm, compression and storage of the fingerprint images, exploitation for real time characteristics extraction and special circuit for matching.The research presented in the thesis mainly includes algorithm research on pre-processing, feature extraction and feature matching of fingerprint images, and the hardware realization of an automatic fingerprint identification system based on the algorithm we developed.The highlights of our research work are the following: (1) The pretreatment to the image to make the process that is going to be done easier. (2) Application of " genetic algorithm "strategy to fingerprint image matching, allowing to be more accurate and feasible, introduced by relative translation and rotation between different images of a same fingerprint. A successful feature matching is thus achieved. (3) Design and implementation of hardware realization of the fingerprint identification system. On the basis of our self-owned algorithms, a hardware platform with DSP as core chip, combined with PLD technique, has been constructed.There are 5 chapters in all.Chapter1: Overview, an introduction to the basic knowledge of the technology, the history of its development and the status inward and abroad.Chapter2: an introduction to the general composition, the function of each. Chapter3: A detailed explanation of the specific political method In this thesis, the focus is on the mode of to realize the fingerprint recognition, The original fingerprint directly from sensor can net be used to gain the characteristic point, So we need to preprocess the fingerprint image in order to get better characteristic.The pre-processing is composed of several steps. First of all, the removement of noise from the fingerprint image is to be done. The reason is that the image will be interfered by various noises. In addition, the fingerprint images are stored in the computer in form of matrix after A/D,D/A transformation, so the noise is unavoidable in the process of digitalization and quantification .Usually, these noises are unrelated to time, so they contain high-level space frequency register. All of those show us the explaination of the necessity of removing the noise before characteristic extraction. The task of smoothing is to get rid of jamming noise and minimize image distortion. To some degree, smoothing reduces the influence caused by the noise on the f...
|
PDF Full Text Request