Abstract

---

A fundamental problem in the field of off-line signature verification is the lack of a signature representation based on shape descriptors and pertinent features. The main difficulty lies in the local variability of the writing trace of the signature which is closely related to the identity of human beings. Hand-written signature is widely used for authentication and identification of individual. The proposed algorithm uses clustering technique for extraction of feature points from the image of the signature. To avoid interpersonal variation 6 signature images of the same person are taken and feature points are trained. These trained feature points are compared with the test signature images and based on a specific threshold, the signature is declared original or forgery.

Introduction

---

The authentication of a person’s identity with the help of his signature increased considerably during the end of the last century. We would like to know who signed a contract or a check whether by the authentic person or not by a criminal. This paper provides an easy to use application that detects digital signatures. 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 processing of individual identification faster and more accurately, the design of an automatic signature system faces a real challenge [1, 2].
Signature verification is different than the character recognition, because signature is often unreadable, and it is just an image with some particular curves that represent the writing style of the person. Signature is a special case of handwriting and is just a symbol. So it is wisdom and necessary to deal with a signature as a complete image with special distribution of pixels and representing a particular writing style and not as a collection of letters and words [2].
The algorithm applied for off-line signature verification is typically a feature extraction method. In this method features are extracted and adjusted from the original signatures. These extracted features are used to distinguish between original and forgery signatures. Two methods have been implemented for extraction of features, namely; clustering and geometrical feature extraction.
The rest of the paper is structured as follows. In sections 2 and 3, overviews of signature verification systems and forgery detection systems are presented. In section 4, methodolocal details are summarized. Sections 5, discusses various implementation details. Finally, the results and discussions are drawn in section 6.

Overview of signature verification systems

---

For centuries, handwritten signatures have been an integral part of business transactions, contracts and agreements. The distinctiveness of a handwritten signature helps to prove the identity of the signer, while the act of signing a document represents the signer's acceptance of its terms and also codifies the document's contents as being official and complete at the time it was signed [3].
Handwritten signature verification can be divided into on-line and off-line verification. On-line verification refers to a process that the signer uses a special pen called a stylus to create his or her signature, producing the pen location, speeds and pressures. While off-line verification just deals with signature images acquired by a scanner or a digital camera. In an off-line signature verification system, a signature is acquired as an image. This image represents a personal style of human handwriting, extensively described by the graphometry. In such a system the objective is to detect different types of forgeries, which are related to intra and inter-personal variability. The system applied should be able to overlook inter-personal variability and mark these as original and should be able to detect intra-personal variability and mark them as forgeries [4, 5].
As compared to on-line signature verification systems, off-line systems are difficult to design as many desirable characteristics such as the order of strokes, the velocity and other dynamic information are not available in the off-line case. The verification process has to fully rely on the features that can be extracted from the trace of the static signature image only. Although difficult to design, off-line signature verification is crucial for determining the writer identification as most of the financial transactions in present times are still carried out on paper. Therefore, it becomes all the more essential to verify a signature for its authenticity. The design of any signature verification system generally requires the solution of five sub-problems: data acquisition, pre-processing, feature extraction, comparison process and performance evaluation [3, 5].

Overview of forgery detection systems

---

Automatic examinations of questioned signatures were introduced in the late 1960s with the advent of computers. As computer systems became more powerful and more affordable, designing an automatic forgery detection system became an active research subject. Most of the work in off-line forgery detection, however, has been on random or simple forgeries and less on skilled or simulated forgeries. Before looking into the landmark contributions in the area of forgery detection, we first enumerate the types of forgeries [3].
Types of forgeries
There are three different types of forgeries to take into account. The forgeries involved in handwritten signatures have been categorized based on their characteristic features [6]. Each type of forgery requires different types of verification approach. Different types of forgeries and their variation from original signature are shown in Fig. 1. We have also attempted to classify the various kinds of forgeries into the following types:
1. Random forgery—The signer uses the name of the victim in his own style to create a forgery known as the simple forgery or random forgery. This forgery accounts for the majority of the forgery cases although they are very easy to detect even by the naked eye [7].
2. Unskilled forgery—The signer imitates the signature in his own style without any knowledge of the spelling and does not have any prior experience. The imitation is preceded by observing the signature closely for a while.
3. Skilled forgery—Undoubtedly the most difficult of all forgeries is created by professional impostors or persons who have experience in copying the signature. For achieving this one could either trace or imitate the signature by hard way.
Fig 1: Different types of forgeries
Ammar et al. [8] have worked on the detection of skilled forgeries. They have calculated the statistics of dark pixels and used them to identify changes in the global flow of the writing. The later work of Ammar [9] is based on reference patterns, namely the horizontal and vertical positions of the signature image. The projections of the questioned signature and the reference are compared using Euclidean distance. Guo et al. [10] have presented an algorithm for the detection of skilled forgeries based on a local correspondence between a questioned signature and a model obtained a priori.

Methodology

---

Clustering
Clustering is basically used in radial basis function neural network (RBFNN) for auto up adjustment of centres of the network to get efficient output. We applied clustering for extraction of features from the original signature image. The obtained adjusted centres are considered as feature points of the signature image. In this method of feature extraction, all the co-ordinates of the signature are taken as input of the network. Equally spaced feature points are taken at random. The random feature points are adjusted using the Eqn. (1)  (1)

where,  Iteration number
t – Centre or feature point
k – Iteration number
n – Centre number.
 – Centre adjustment parameter (step size)
x (i) –  ith  input
This actually tries to move the centre closest to the input towards the input according to a pre-defined step-size. After a few iterations the centres adjust themselves completely and can be considered as the feature points of the signature image. These feature points are mapped on to the signature image to be tested. A threshold is decided as the rule to distinguish between original and forgery signatures. Based on that specific threshold it is decided whether the signature is a forgery or the original one [2].
Geometrical feature extraction
The signature image is divided into several sub-images or blocks, like blocks containing equal number of pixels (signature pixels generally black pixels), equi-spaced-equal-sized blocks etc. Then the centroid of each of these blocks or sub-images acts as a feature points of the image.
Image binarization
The input image is taken and converted into gray-scale image which is further pre-processed i.e. it is binarized. The binary image of the signature contains only 0’s and 1’s. Where 0’s represents the signature boundary and 1’s represents the blank white area or the background region. This is done by specifying a specific threshold, above which every gray value is 1 and below which every gray value is 0.

Implementation

---

Firstly, sample signatures are collected and kept in the database. Here for every person n signature samples are collected for database. For verification test signatures are collected against the sample signatures. These test signatures have to be verify if it is genuine or forgery. Each of the signatures (samples and corresponding test) has to take within a same sized rectangular area on paper by pen and collect the image of that rectangular area. The signed paper is then scanned by a gray scale scanner. Therefore, the signature image will be processed in four stages in order to determine if it is genuine or forgery.
The system deals with the static scanned image of the signatures. Unwanted images i.e. noises may include during the scanning process of the signature. The pre-processing of the signature images is related to the removal of noises and thinning. The goal of thinning is to eliminate the thickness differences of pen by making the image one pixel thick. To remove noises, the images are pre-processed by filtering techniques. For thinning morphological operations can be applied. Pre-processing is required for both sample and test signature images [11].
Proposed Algorithm: The algorithm has following steps:
i) Firstly, all input of the original signature image is converted into binary image and then image is skeltenize.
ii) Find the locations of all the black pixels (signature pixels) from the background image (white). Assign equally spaced centres to the signature image.
iii) By taking x-y co-ordinates of the image as input adjust the centres using Eqn. (1) to the image signature. Adjust them for a few training signature images. Take the finally adjusted centres as the feature points of the signature. Create a distance matrix from feature points.
iv) Scale the distances by a factor of 100 to make it rotational and scale invariant.
v) Take the input signature image and convert it into binary image.
vi) Extract the features as described above and create a distance matrix.
vii) Scale these distances by a factor of 100 and map this distance matrix with the feature point distance matrix.
viii) The number of errors less than 5.0 should be above 18 and total error is less than 155.
ix) If the test signature matches both the above criteria then the signature is declared original else it is forgery.

Results and Discussion

---

The algorithm is implemented on personal computer (1.8GHz CPU, 2GB RAM). It is a well known fact that any automatic signature verification system requires a very small training set of signatures. For this reason, we have set the number of training signatures for each individual at six. The algorithm consists of two distinct divisions. First features from the sample signature image and test signature image have to be computed. Then a decision will be made based on the clustering results between the computed feature of sample signature image and test signature image.
A small database of off-line signatures was prepared. Each of 4 individuals contributed one signature thereby creating 4 genuine signatures. Some were asked to forge four other writers' signatures, eight times per subject, thus creating 32 forgeries. One example of training and test signatures are shown in shown in Fig. 2 and Fig. 3. The logic for classification of a test signature is also derived experimentally considering the simulated outputs of all the networks all together. For a typical test signature simulated output all the network may not agree with same result. Logic is developed prioritizing the networks for their outputs so that the trade off between false acceptance rate (FAR) and false rejection rate (FRR) is taken in care. We found that all the feature set can uniquely classify a signature. But the FAR and FRR obtained from different feature set are different. It is found that FAR is smaller than FRR for features obtained from vertical sectioning of the signature and FRR is smaller than FAR for features obtained from horizontal sectioning of the signature. From simulation results the number of errors below 5.0 is 22, 29 and 30. Hence test signatures are declared forgery and it is shown in Fig. 4(a) and Fig. 4(b).
Fig. 2: Some training signatures
Fig. 3:   Test signatures
Fig. 4: Test signatures for random forgery and simple forgery.
This paper presents a method for off line hand written signature verification with higher accuracy. In this paper we have introduced a procedure to extract features from handwritten signature images and computed feature is used for verification. A novel off-line signature verification algorithm has been presented which uses the soft-computing clustering technique. As clear from the simulation results the algorithm is capable of verifying almost all the signatures. The equi-spaced features are adjusted or updated using the cluster update algorithm and these centres or feature points are trained using the training signatures in the database to avoid interpersonal and intrapersonal errors as much as possible. Despite our best efforts still there are some loop holes in the algorithm, due to which there are some errors in the result. The algorithm has its pre-specified threshold of max error = 155, pre-specified step size in the clustering that is 0.005 etc. All these parameters can be made adaptive, that will adjust them according to the input given to them. The extraction of exact signature from the signature image sometimes produces error in verifying the signature. This approach works well if there is a high variation in the original signature, but for signatures with low variation, it produces incorrect results.

References

---

1. A.Ismail, M. Hajjar and M. Quafafou, “Off-line signature verification uses single and parallel processing”, in Proc. 2009 Intl Conf: Sciences of Electronic, Technologies of Information and Telecommunications, 2009, pp. 1-6.
2. S.Biswas, T-H Kim and D. Bhattacharyya, “Features extraction and verification of signature image using clustering technique”, International Journal of Smart Home, vol.4, no.3, pp. 43-55, 2010.
3. M.Hanmandlu, M. Hafizuddin M. Yusof and V. K. Madasu, “Off-line signature verification and forgery detection using fuzzy modelling”, Pattern Recogn, vol. 38,pp. 341-356, 2005.
4. R.Plamondon, “The design of an on-line signature verification system: from theory to practice”, Int. J. Pattern Recogn. Artif. Intell., vol. 8,  pp. 795-811, 1994.
5. M.Ammar, Y. Yoshida and T. Fukumura, “A new effective approach for off-line verification of signatures by using pressure features”, in: Proc. Intl Conf on Pattern Recogn, 1986, pp. 566-569.
6. R.Plamondon and G. Lorette, “Automatic signature verification and writer identification: the state of the art”, Pattern Recogn, vol. 22, Issue 2, pp. 107-131, 1989.
7. A.S.Osborn, Questioned Documents, Patterson Smith Publication Corporation, 1973.
8. M. Ammar, Y. Yoshida and T. Fukumura, “A new effective approach for off-line verification of signatures by using pressure features”, in: Proc. Intl Conf. on Pattern Recogn., 1986, pp. 566-569.
9. M.Ammar, “Progress in verification of skilfully simulated handwritten signatures”, Int. J. Pattern Recogn. and Artif. Intell., vol. 5, pp. 337-351, 1991.
10. J.K.Guo and D. Doermann, “A. Rosenfeld, Off-line skilled forgery detection using stroke and sub-stroke properties”, in: Proc. 2000 Intl Conf on Pattern Recogn, 2000, pp. 355-358.
11. R.C.Gonzalez, R. E. Woods “Digital image processing”, 2nd edition, 2003.

Source(s) of Funding

---

none

Competing Interests

---

none

Disclaimer

---

This article has been downloaded from WebmedCentral. With our unique author driven post publication peer review, contents posted on this web portal do not undergo any prepublication peer or editorial review. It is completely the responsibility of the authors to ensure not only scientific and ethical standards of the manuscript but also its grammatical accuracy. Authors must ensure that they obtain all the necessary permissions before submitting any information that requires obtaining a consent or approval from a third party. Authors should also ensure not to submit any information which they do not have the copyright of or of which they have transferred the copyrights to a third party.
Contents on WebmedCentral are purely for biomedical researchers and scientists. They are not meant to cater to the needs of an individual patient. The web portal or any content(s) therein is neither designed to support, nor replace, the relationship that exists between a patient/site visitor and his/her physician. Your use of the WebmedCentral site and its contents is entirely at your own risk. We do not take any responsibility for any harm that you may suffer or inflict on a third person by following the contents of this website.