A robust and an efficient offline signature verification algorithm

A robust and an efficient offline signature verification algorithm

An offline signature verification system is proposed in this paper. The proposed model has two stages:
preprocessing and eigen-signature construction. In the preprocessing stage, we convert a scanned signature to a shape form and eigen-signature construction is proposed for extracting the feature vector from a shape formed signature. Experiments have been conducted on the newly created Kannada offline signature database to exhibit the performance of the proposed model. A comparative analysis is provided with the recently proposed texture features based offline signature verification system on the
publicly available gray signature database to exhibit the performance of the proposed model. A robust and an efficient offline signature verification algorithm

We have seen the usage of biometrics based security/verification system in recent years to prevent
unauthorized access to all kinds of data either in electronic (softcopy) form or in document (hardcopy) form. There is concerned effort to replace traditional means of identification such as passwords by biometric based authentication systems. A handwritten signature, being a behavioral biometrics, is well accepted socially and legally as a convenient means of authorization and identification. The need to guarantee the authenticity of each document remains urgent, and demands more efficient, controlled and reliable method of signature verification. Existing signature verification systems are generally
classified either online or offline approaches, depending on the nature of data acquisition and application involved. Online signature verification systems generally present a better performance than the offline signatures verification systems. The velocity, acceleration, direction of pen movement, pressure and forces are some of the features used in many of the existing online signature verification systems [2, 10, 12, 13]. However online signature verification system necessitates the presence of the signer at the time of both acquiring the reference signature and the verification process which is not
welcome by many applications. Hence off-line verification methods have more practical application areas than that of the online signature verification methods

the features or strokes of the signature patterns and build a statistics of these variations from the training set. Guo et al. [16] addressed the problem by establishing a local correspondence between a model and a questioned signature. Hung and Yan [17] present a method using a combination of
static image pixel features and pseudo-dynamic structural features. Fang et al [20] proposed a method dealing with the smoothness between genuine signatures and forgery ones. Here the smoothness index is combined with other useful global geometric features and used for offline signature verification. Deng [22] justifies that wavelet approach is the best method, used in an offline signature verification system,
as spectrum based approach. Parizeau and Plamondon [23] worked on Neural Network approach that has been well suited for reasonable performance in offline signature verification problem as pattern recognition problem. Sabourin and Drouhard [19] proposed directional probability density
functions and a completely connected feed-forward neural network classifier to build the first stage of a complete automatic handwritten signature verification system. Hanmandlu [11] proposed the verification and forgery detection system based on the fuzzy model, where as Piyush and Rajagopalan [21] and I.Guler et al., [24] has experimented with the dynamic time warping. Thus offline signature
verification will lead future trend for personal identification because of its multiple advantages such as gray scale based representation and spatio-luminance analysis for feature extraction. In addition, it does not require any special processing devices and has more practical applications. But preprocessing is more complex and time consuming in offline systems due to unavailability of the dynamic information. Devising an efficient and accurate offline signature verification system is a challenging task as
signatures are sensitive to geometric transformations, inter- personal signature verification