Research On The Key Algorithms Of Automatic Fingerprint Identification System

Because human fingerprint is permanent and unique, fingerprint verification has been widely used in many fields. Today, with the development of science and technology, the Automatic Fingerprint Identification System(AFIS) gets the rapidly development and application. However, as fingerprint verification algorithm is complicated, its processing usually demands high computer hardware and the high cost has affected the application development.In this thesis, we make a widely study of the algorithms which are popular used nowaday, and present a better algorithm by the optimization research of the algorithms. A set of complete fingerprint verification algorithm include several steps, namely , image enhangcement preprocessing, binaryzation, thining, the minutiae extration and fingerprint matching. Nowaday, there are a lot of algorithms averable in each steps. But, considering the using based on the embody application, this algorithms have defects in many affects. The traditional fingerprint image segmentation usually use the image gray mean, gray-scale variance and direction information, but simple threshold can't carry out the accurate and effective segmentation because of all kind of complicated image. So we carefully analysis the texture of the image and present a new character describe the texture information from the frequency domain. At the same time, we designed a lineary classifier for the image segmentation. Moreover, through the minutiae extration, we get lots of false minutiae. We introdue a suitable algorithm to reduce the num of false minutiae. Today, there are many algorithms about the minutiae matching. They are all much complicated. Here, we will propose a triangle-based fingerprint minutiae matching algorithm. In this thesis, the main contributions are as follows:(1) We present a comprehensive preview about fingerprint image processing on the state of the core algorithm related to AFIS.(2) We introduce a new fingerprint image segmentation algorithm. Then we use a lineary classifier to perform the image segmentation. In addition, We also present a segmentation algorithm used gray variance information. (3) The image binaryzation is an important step for the performance of the automatic fingerprint recognition. We propose the adaptive threshold-dynamic binarization method which is widely used in this field. Furthermore,We use the mathematics morphology to process the image, so we can abtain good result.(4) Because of the noise interference, the minutiaes we obtained include many false minutiaes. So a sort of suitable remove algorithm is necessary. We carefully analyze the reason of bringingfalse minutiaes and introduce the processing method.The research project has done a optimization research on fingerprint verification algorithm which is to design a set of rapid and effective mothed used in the embedded-application fingerprint process system. We have done some works and got a few achievement, but we still need to research deeply to abtain better the effect of AFIS.