Font Size: a A A

Offline Chinese Signature Verification System Research And Implementation

Posted on:2009-12-01Degree:MasterType:Thesis
Country:ChinaCandidate:Z H ZhengFull Text:PDF
GTID:2178360242483905Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
With the rapid development of information technology, a reliable, fast and practical technology to identify identity is required in many applications. Signature is one of the widely used biological characteristics in personal identification. The Chinese offline signature verification technology has become a hotspot.The basic idea of offline Chinese signature verification is that: firstly, do some preprocessing to the original signature image; secondly, extract the image features; and finally, the similarity of feature vectors of input signature image with the feature vectors in the database is calculated to determine the true signature. In this paper, the key technologies of signature verification are studied.Some attentions were spent on the preprocessing technology in the signature verification. In this paper different preprocessing steps are made to different type of features. For static features, the preprocessing steps are: mean filter - binarization - salt and pepper filter - refinement; for texture features, the preprocessing steps are: Mean filter - binarization - salt and pepper filter– edge extraction; for pseudo-dynamic features, the preprocessing steps are: mean filter - binarization - salt and pepper filter– the extraction of foreground gray image.The algorithms of feature extraction in the signature verification are studied. The shape feature, texture feature and pseudo-dynamic feature are used in this paper. Lots of experiments were done to test the performance of these features.In the decision-making stage, using a variety of classifier verify signatures, under the conditions of the different characteristics, get each optimal classification from the characteristics, which be used in the program of features integration.Threshold: because in the same indicators, including FAR and FRR, the same characteristics, the same categories and the different signatures, the threshold is different, the system uses dynamic threshold, that is different for different signatures, characteristics and categories.System frame: inserts pattern recognition system in disguised form in the pattern recognition system's training stage, in order to seek the dynamic threshold.In the above theoretical research, signature verification system is designed, using Visual C++ 6.0 tool.
Keywords/Search Tags:signature verification, signature template, dynamic threshold, category distance
PDF Full Text Request
Related items