Signature Verification System Based on Support Vector Machine Classifier

  Sonal Borase,  Dr.Javed Rasheed Shaikh

The paper presents associate degree off-line signature verification system exploitation support vector machine technique. world options ar extracted from the signatures exploitation argonon rework. for every registered user within the system info variety of reference signatures ar listed and aligned for statistics info extraction concerning his signature. Dynamic time warp rule is employed to align 2 signatures. throughout support vector machine classifier coaching, variety of real and solid signatures ar chosen. A check signature’s verification is established by initial positioning it with every reference signature for the claimed user. The signature is then classified as real or forgery, in keeping with the alignment scores that ar normalized by reference statistics, exploitation commonplace pattern classification techniques. employing a info of 2250 signatures (genuine signatures and good forgeries) from seventy five writers within the planned signature verification system a performance of roughly eighty two is achieved.

Keywords- signature verification; support vector machine; radon transform; dynamic time warping

Unique Identification Number - IJEDR1904035
Page Number(s) - 202-207
Pubished in - Volume 7 | Issue 4 | October 2019
DOI (Digital Object Identifier) -   
Publisher - IJEDR (ISSN - 2321-9939)

  Sonal Borase,  Dr.Javed Rasheed Shaikh,   "Signature Verification System Based on Support Vector Machine Classifier", International Journal of Engineering Development and Research (IJEDR), ISSN:2321-9939, Volume.7, Issue 4, pp.202-207, October 2019, Available at :http://www.ijedr.org/papers/IJEDR1904035.pdf