Font Size: a A A

Personal Identification Based On Multimodal-Biometrics

Posted on:2007-08-25Degree:MasterType:Thesis
Country:ChinaCandidate:B SunFull Text:PDF
GTID:2178360212467035Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
According with the development of biometrics, multimodal-biometrics was greatly developed. In this paper, what we do focus on a multi-biometrics system based on face and palm-print. We work to optimize the combination of multi-classifier and improve the performance the multi-biometrics system. Here is a summary of my work:(1) A novel idea and framework are presented to implement the kernel methods on high-dimensional data. We implement this method to speed up feature extraction of face and palm-print in the multi-biometrics system. In order to extract the feature effectively, we use kernel methods to extract feature of face and palm-print. But both face image and palm-print are high dimensional data, if we use kernel method to extract feature of them directly, it will correspond to a high computational cost. Further more, if we combine the two feature extraction process together, efficiency will be the key. In this paper, we propose a novel method to implement kernel method on high dimensional data. A remarkable character of the framework is that there are two feature extraction processes. The first feature extraction process is performed to transform high dimensional samples into low dimensional ones. And, the second feature extraction process is implemented based on the obtained low dimensional data. With the novel framework, the kernel methods will become much efficient. Moreover, all the kernel methods can work with the framework. The experiments on face images show the validity of this framework. Moreover, they can achieve accuracies which are not lower than those associated with the corresponding naive kernel methods.(2) We propose a new fusion method that based on details of output-vector of single classifier. This method focus on the measurement level output information of each classifier, it convert the output distance to posterior possibility, then we can give the weight face and palm-print according to this, the weight value can be expressed as the probability that the given sample belong to every possible class. At last, we use the Bayes theory to fusion the result. This fusion method is easy to understand. Compared with...
Keywords/Search Tags:multi-biometrics, multi-classifier fusion, face recognition, palm recognition, weighted method
PDF Full Text Request
Related items