Font Size: a A A

Fusion Of Face And Palmprint On Decision Level For Identity Verification

Posted on:2007-03-10Degree:MasterType:Thesis
Country:ChinaCandidate:N NiFull Text:PDF
GTID:2178360212967021Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Biometrics is a technology that uses human characteristics or behavioral traits to verify person identification by computer automatically. It involves some konwledge about pattern recognition, artificial intelligence, image processing and other subjects. In the recent years, single modal biometric system is not satisfactory for practical use, so the multi-modal biometric system is becoming a hot issue. In this thesis, some research on the combination of face and palmprint on the decision level is done. The thesis includes the following parts:(1) Many approachs for combining on the decision level of the multi-modal biometric system is summarized and compared, some experiments is done to test the performance of different classifiers.(2) Research on the performance in the case of single training sample problem, the way of traditional linear weight combination is improved according to the specific case of single training sample, and the result of experiment shows that the method is valid.(3) Use wavelet transform to extract the feature of face and palmprint image, a combination algorithm based on evolutionary strategy is brought forward. This approach can calculate proper weight for each classifier and improved the accuracy. And this approach is easy to realize, can use much less time than most of other combination algorithms.(4) Some improvements on the integration method used in character recognize are presented. This improved method is used in face and palmprint combination, and is proved to be effective and efficient.The methods brought forward in this thesis are effective and can solve some problems in multi-modal fusion of biometrics. The research on classifier combination is not only useful and helpful to biometrics field , but also to other fields in pattern recognition.
Keywords/Search Tags:biometrics, multi-modal fusion, classifier combination, evolutionary strategy, integration decision
PDF Full Text Request
Related items