DEVELOPMENT OF OFFLINE SIGNATURE IDENTIFICATION SYSTEM

Abstract

Signatures are special case of handwritten characters. So it is better to take this as an image. So this makes it even more difficult to analyze it. Depending upon nature of data acquisition and application involved, signature identification system can be classified into two parts:-online signature identification system and offline signature identification system. In this theses report five methods for offline signature identification system are used and there result are analysis in this chapter. In first method, chain code direction feature and binary intensity of signature is used. In this method, first signature is divided into 100 segment parts. Then chain code direction feature and binary intensity is calculated for every part. Thus there are total 500 (400 + 100) features, 5 in one segment • part of signature. Thus no. of features is very high, so to reduce the features, all features in one segment of signature image is added. Thus it reduces the no. of features to 100 features. However even 100 feature for one signature is very high for neural network. So these features are further reduces to 50 per signature. This is done by taking top 50 signature of a signature image. Now with these 50 features per signature is trained by ANN and results are obtained. In second method, some global features with moment invariant vectors are used to identify a signature. In this method 8 global properties like height to width ratio, max vertical and horizontal projections, vertical and horizontal centers, image area, max vertical and horizontal projection points are used. Seven moment invariant vectors are calculated. Thus there are 15 features per signature used. Now these features are used to train ANN. In third, fourth and fifth method, identification of signature is done in two steps. In first step, signature is recognized and in second step it is verified, i.e. it is right or fake. These methods depend upon types of features used to train ANN. In first step only 1 ANN is used for recognition, but in second step separate ANNs are used for individual user to verify.