The Research Of Offline Identification In Handwriting Signature Based On The Lagrangian Support Vector Machine

The popularity of computer networks has made the exchange between the people more and more convenient and frequent, but the subsequent security problem has become increasingly prominent. Traditional identification methods,such as passwords, keys and the like, have not ever met the needs of the society. The development of biometric identification technology provides a more convenient and reliable solution for us,such as iris,fingerprint and so on. Signature as a behavioral characteristic, compared to other biometric identification, has the advantage of easier access and being able to share, which is one of the most widely used biometric identification. The handwriting recognition is divided into two types:offline identification and online identification. The paper mainly studies the methods of offline identification of handwriting signature.The paper uses the signature image as the target object, in-depth studies the preprocessing technology, feature extraction, selection techniques and identification technology in the handwriting technology.In the study of preprocessing of signature image techniques, by filtering, bianizating, thinning and normalizating the original signature image, the study has laid a good foundation for the subsequent feature extraction.During the feature extraction process, the paper extracted from a handwritten signature five static characteristics and4peseudo-dynamic characteristics of9characterized. After feature extraction, using the law of probability distance Bhattacharyya nine characterized selected finally obtained the seven characteristic value for the judgment of the recognition process characterized.In the comparison discernment process, the paper describes the main structure of the standard support machine(SVM) and classification principles, analysis of the problems of standard support vector machine(LSVM)the proposed Lagrangian support vector machine to identify and applied off-line recognition of handwritten signature. It is easier to implement the LSCM algorithm than the standard SVM algorithm, and the speed is more accelerated recognition, and the effect is more ideal to obtain a higher recognition rate.Finally, the paper designed offline recognition system based on Lagrangian support vector machine method using MATLAB simulation algorithm, it has better recognition rate.