Font Size: a A A

Design Of Fingerprint Recognition Based On Opposite Direction Error Spread

Posted on:2006-12-07Degree:MasterType:Thesis
Country:ChinaCandidate:X L YangFull Text:PDF
GTID:2168360155953308Subject:Measuring and Testing Technology and Instruments
Abstract/Summary:PDF Full Text Request
Fingerprint feature is one of the unchangeable features in one's life , and the possibility that fingerprint feature of different people are the same is almost zero .It is because the fingerprints possess the unique sign that the application of the fingerprint is put on an important position by people. An automatic recognition system of fingerprint contains three stages generally : the pretreatment of picture of fingerprint , feature abstraction and matching recognition. In overseas country , automatic recognition of fingerprint mostly utilize the direction of fingerprint line ,and it doesn't have the relation with fingerprint status point . It is a principal way which was used early stage by fingerprint automatic recognition. At home, there is a new fingerprint detail status about the relational position that distinguish the automatic checking method. But fingerprint image is usually needed to be interfered the position by people. At present, fingerprint automatic recognition has been widely applied into the public security organ and some enterprises. Owing to the high level of computer technology , utilizing fingerprint automatic recognition to discern criminals and uncovering has become a necessary means and method concerning the public security organ . But we must depend completely on our technology of fingerprint system of automatic recognition which contains the following three shortcomings:the inclined low rate of recognition , the inclined slow speed and the wasting time, and fingerprint experts need to interfere largely, and internal fingerprint recognition is all limited into the recognition technology of traditional mode ,usually containing statistical and structural mode ,recognition the two methods. It is difficult to see from some outside public materials of these system :with regard to fingerprint recognition system of statistical mode ,adopting the analysition of fingerprint line ,utilizing feature mode and establishing mode containing the length and direction of fingerprint line. With regard to fingerprint system of structural mode recognition , from the local structure fracture of fingerprint ,analyzing the distribution of fingerprint feature point is established relational feature direction . Traditional mode-place of matching method is one of the main ways that resolve image matching .Containing direct relational matching method , similar checking method , Hausdorff distance method ,unchangeful square method , shape special point method and so on .Among them some calculation methods are simple ,but some calculation quantities are extremely larger ,or matching effect is not excellent ,or matching feature of needing is difficult to gain in use; Some calculation methods are precise , but those need disposion in advance, and it can't meet the requirements of speed. With regard to the automatic recognition system of traditional fingerprint has limitations. Many experts of fingerprint automatic recognition technology set different views of improving .Now, there is a tendency, experts want to discard the matching method of traditional mode-plate, adopting directory wave to draw fingerprint feature , then proceeding the matching and recognition of mode-plate. The writings make use of the artificial nerve net (ANN)that possesses many merits the parallel disposition and studying oneself of the problem of unline shape . So , nerve net tales widely application and conspicuous effects , in some fields , for example , in mode recognition, image disposition , signal disposion and automatic controlling and so on. We try to take together the artificial nerve net (ANN) and the automatic recognition system of traditional fingerprint recognition to make the fingerprint recognition more intellectually ,improving thespeed and rate recognition . The article itself set the structural mode matching calculation method of fingerprint ,and the method successfully resolved the problem of recognition kinds of defect and abnormal fingerprint .Setting a more special calculation method of fingerprint image ,and the method depends on the gray degree of image points and the effect is good , overcoming the phenomenon that the calculation method greatly lost fingerprint line . Meanwhile , the stage of beforehand disposition successfully designed kinds of directory filter ,and the effect of filter is excellent . The author of the article sit s automatic recognition calculation method of fingerprint , taking the whole distribution of fingerprint feature and partial feature distribution together , to a certain extent imitating the thought of experts of fingerprint recognition ,the high degree of systematic intellecting . The article adpting the method of structural mode takes the following several characteristics : (1) Flexible selecting of feature arrow number . While the sample storehouse of fingerprint is short , the drawing of n may be too small that attains the precise recognition; while the fingerprint storehouse is too large , the drawing of n must be high to distinguish every fingerprint precisely . (2) The excellent feature of calculation method of system. Provided the value of n is high , even if the drawing of K is high to control the zero of the fault recognition rate , system about the complete map of fingerprint takes the high rate of recognition. Because of adopting the structural feature to recognize and match , system about tolerate degree of changing and revolving of the map of recognition fingerprint is too excellent . (3) About the limitation of the BP method, we set the two improving measure both adding the divisor γp and spring reading , the experiment proving that these...
Keywords/Search Tags:fingerprint recognition, feature extract, directional image filter, opposite direction error
PDF Full Text Request
Related items