Handwritten Signature Recognition System Using Euler ... - CiteSeerX

International Journal of Advanced Information Science and Technology (IJAIST) Vol.25, No.25, May 2014

ISSN: 2319:2682

Handwritten Signature Recognition System Using Euler Number Souvik Chatterjee 2 nd year M.Tech IT Courseware Engg. School of Education Technology Jadavpur University Kolkata, India [email protected]

Mr. Joydeep Mukherjee Asst. Professor, School of Education Technology Jadavpur University Kolkata, India [email protected]

Abstract : This paper reports the design, implementation, and evaluation of a research work for developing an digital handwritten signature identification system using binary image analysis. The developed signature identification system mainly used binary image analysis provided by MATLAB environment. In order to train and test the developed signature identification system, an in-house hand signatures database is created, which contains hand signatures of 5 persons (2 males and 3 females) each of which is repeated 10 times. Therefore, a total of 50 hand signatures are collected. The collected hand signatures have gone through pre-processing steps such as producing a digitized version of the signatures using a scanner, converting input images type to a standard binary images type, cropping, normalizing images size, and reshaping in order to produce a ready-to-use hand signatures database for training and testing the signature identification system. Feature such as EULER NUMBER is then selected to be used in the system, which reflects information about the structure of the hand signature image. Overall, the handwritten signatures based system obtained an average recognition rate of 80% for all persons.

processing of individual identification faster and more accurately, the design of an automatic signature verification system faces a real challenge. Recognition can be performed either Offline or Online based on the application. Online systems use dynamic information of a signature captured at the time the signature is made. Offline systems work on the scanned image of a signature. In this paper we present a method for Offline recognition and verification signatures using analysis of that collected binary image.

Keywords : Signature Recognition, Binary image analysis, Euler Number 1.

Introduction

Handwritten signature is one of the most widely accepted personal attributes for identity verification. As a symbol of consent and authorization, especially in the prevalence of credit cards and bank cheques, handwritten signature has long been the target of fraudulence. Therefore, with the growing demand for

2.

Related Work

Ismail et al. [1] proposed an off-line Arabic signature recognition and verification technique. In their paper, a system of two separate phases for signature recognition and verification is developed. The recognition phase, some features based on Translation, circularity feature, image enhancement, partial histogram, centers of gravity, global baseline, thinning etc. are extracted. A set of signature data, consisting of 220 genuine samples and 110 forged samples is used for experimentation. They obtained a 95.0% recognition rate. Ozgündüz et al. [2] have also presented a signature recognition and verification system by using SVM. Binarization, noise reduction, width normalization and skeletonization have been done as preprocessing section. Then, global, oriented and grid features have been extracted. The global features include area, width and length ratio,

16


vertical and horizontal central gravity, etc. Oriented features calculate the signature line gradient. The signature image is divided to 60 equal parts and the signature area is gained in each one. SVM classifier has been used to recognize and verify signature. The signature recognition error has been reported 5%. Kaewkongka, Chamnongthai and Thipakom [3] proposed a method of off-line signature recognition by using Hough transform to detect stroke lines from signature image. The Hough Transform was used to extract the parameterized Hough space from signature skeleton as unique characteristic feature of signatures. In the experiment, the Back Propagation trained Neural Network was used as a tool to evaluate the performance of the proposed method. The system was tested by 70 test signatures from different persons. The experimental results reveal the recognition rate 95.24%. Radmehr, Anisheh and Yousefian [4] proposed a new offline signature recognition is presented. In the first step, radon transform is initially applied on the signature image with 0°, 45°, 90° and 135° angles. Then fractal dimension of the obtained vectors is calculated and the results are fed into SVM classifier. The simulated results indicate that the proposed method has a high accuracy in signature recognition. In the last few decades, many approaches have been developed in the pattern recognition area, which approached the offline signature verification problem.

3.

System Design & Architecture

The architecture of the handwritten signature identification system using handwritten signatures is divided into two main phases. During the first phase, all training steps are performed, whereas during the second phase of the system’s architecture all testing/matching steps are performed. Following figure shows the main architecture of the system. In order to use the system, hand signature images database must be collected. Details on the hand signature images database are discussed in the next section.

ISSN: 2319:2682

Architecture of the System

4.

Database

In order to develop the handwritten signatures-based automatic signature identification system, a handwritten signature database is required. An inhouse handwritten signature database is collected, which contains images of 5 persons (2 males and 3 females). Each person was asked to sign on a white sheet that contains a table with 10 cells for 5 times. It is important to highlight that all participants used a black colored (same) pen to sign on the white sheet. Therefore, the total number of handwritten signature images in this handwritten signature database is 50. These 50 images are distributed into training and testing data sets, whereby the training data set contains 5 repetitions per person, whereas the testing data set contains the remaining 5 repetitions per person. Therefore, a total of 25 images are used to train the handwritten signature-based system, whereas a total of 25 images are used to test the system. Once all persons finished signing on the white paper, the paper is scanned using the scanner in order to produce a digitized version of the hand signatures. Each cell is cropped and saved into a separate (.jpg) file and then converted to a binary image using MATLAB command.

17


ISSN: 2319:2682

Number of samples in testing set = 5 Combined accuracy = 80% Plot :

Signature sample before processing

Binary form of that signature

5.

System Development

The training set has 5 classes each having 5 sample signatures. Euler number of those 5 binary images are computed and thereafter average of 5 values is computed. It would work as the feature vector of that class. All the 5 vectors are stored in a array(Features Database). Same process is repeated at the time of testing. Instead of computing their average, each testing value is compared with the mean value of each class in training set. Hence the accuracy is computed on correct recognition.

6.

EXPERIMENTAL RESULTS AND DISCUSSION

The experimental work in this research is evaluated based on the number of correctly identified handwritten signature images. This number is then divided by the total number(number of class) of the testing handwritten signature images, and then multiplied by 100 in order to get the percentage of the accuracy. Number of class = 5 Number of samples in training set = 5

18


7.

Conclusion & Future Work

In this study, we presented an off-line signature recognition and verification system which is based on extract some signature image feature. In the future, this work can be expanded more as new features can be included to recognize image. And also it will the duty of expander to maintain the accuracy even if new features are included. This algorithm is tested on the dataset where the train and test data are slightly changed. This algorithm can be extended on the dataset where the train and test data are vastly changed, not only in expression but also in illumination.

8.

ISSN: 2319:2682

Authors Profile S.Chatterjee received the B.Tech degree in computer science & engineering in 2012. . Currently doing M. Tech in IT Courseware Engineering from Jadavpur University, Kolkata. J. Mukherjee received the M.Tech degree from Jadvapur University. Currently working as an asst. professor in SET, JU, Kolkata.

References

[1] M.A. Ismail, Samia Gad, “Off-line Arabic signature recognition and verification”, Pattern Recognition Society 2000, page 1727-1740 [2] E. Ozgündüz, T. entürk, M.E. Karslıgil, "Off-Line Signature Verification and Recognition by Support Vector Machine". European Signal Proc. Conf. Turkey. Sep 2005. [3] T. Kaewkongka, K. Chamnongthai, B. Thipakom, “Off-Line Signature Recognition using parameterized Hough Transform”, ISSPA Brisbane, Australia, 1999. [4] M.Radmehr, S.M.Anisheh, I.Yousefian, “Offline Signature Recognition using Radon Transform”, World Academy of Science, Engineering and Technology 62 2012. [5] R.M. Haralick, Statistical and structural approaches to textures, Proc. IEEE, vol. 67, 1979, pp. 786 - 804. [6] M. T. Das, and L. C. Dülge, “Off-Line Signature Verification with PSONN Algorithm,” International Symposium on Computer and Information Sciences, Ankara, Turkey, pp.1 – 6, 2007.

19