No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Machine Learning for Signature Verification
Abstract
Signature verification is a common task in forensic document analysis. It is one of determining whether a questioned signature matches known signature samples. From the viewpoint of automating the task it can be viewed as one that involves machine learning from a population of signatures. There are two types of learning to be accomplished. In the first, the training set consists of genuines and forgeries from a general population. In the second there are genuine signatures in a given case. The two learning tasks are called person-independent (or general) learning and person-dependent (or special) learning. General learning is from a population of genuine and forged signatures of several individuals, where the differences between genuines and forgeries across all individuals are learnt. The general learning model allows a questioned signature to be compared to a single genuine signature. In special learning, a person's signature is learnt from multiple samples of only that person's signature- where within-person similarities are learnt. When a sufficient number of samples are available, special learning performs better than general learning (5% higher accuracy). With special learning, verification accuracy increases with the number of samples. An interactive software implementation of signature verification involving both the learning and performance phases is de- scribed.
... This approach is unique in learning authentic signatures without needing forgeries as a reference. The effectiveness of this method is demonstrated on Persian databases (PHBC and UTSig) and Latin databases (MCYT-75 and CEDAR [31]), where it outperformed existing state-of-the-art results. Wei et al. [32] introduce the Inverse Discriminative Networks (IDN) model, designed for writerindependent handwritten signature verification. ...
... This model is notable for incorporating four network streams, each analyzing pairs of signature samples. The IDN model's performance is impressive, showing high verification accuracy on a comprehensive Chinese signature dataset and international datasets such as CEDAR [31], BHSig-B, and BHSig-H [33]. In 2020, Jain et al. [34] developed a shallow Convolutional Neural Network (CNN) approach for verifying handwritten signatures. ...
... Lastly, Liu et al. [43] presents a region-based deep metric learning network for offline signature verification. This method is applied to writer-dependent and writer-independent scenarios, achieving competitive Equal Error Rates (EERs) on challenging datasets like CEDAR [31] and GPDS. These developments collectively highlight the dynamic and evolving nature of signature verification technology. ...
Automated signature verification on bank checks is critical for fraud prevention and ensuring transaction authenticity. This task is challenging due to the coexistence of signatures with other textual and graphical elements on real-world documents. Verification systems must first detect the signature and then validate its authenticity, a dual challenge often overlooked by current datasets and methodologies focusing only on verification. To address this gap, we introduce a novel dataset specifically designed for signature verification on bank checks. This dataset includes a variety of signature styles embedded within typical check elements, providing a realistic testing ground for advanced detection methods. Moreover, we propose a novel approach for writer-independent signature verification using an object detection network. Our detection-based verification method treats genuine and forged signatures as distinct classes within an object detection framework, effectively handling both detection and verification. We employ a DINO-based network augmented with a dilation module to detect and verify signatures on check images simultaneously. Our approach achieves an AP of 99.2 for genuine and 99.4 for forged signatures, a significant improvement over the DINO baseline, which scored 93.1 and 89.3 for genuine and forged signatures, respectively. This improvement highlights our dilation module's effectiveness in reducing both false positives and negatives. Our results demonstrate substantial advancements in detection-based signature verification technology, offering enhanced security and efficiency in financial document processing.
... Machine learning approaches are discussed in [14]. The approach uses statistical learning methods in order to learn to predict the class based on the similarity of the features between the known samples and the new ones. ...
... An exact comparison can be made only with the approaches from Kovari and Charaf [6][5], since they use for evaluation the same dataset as our case study. The other two approaches from [14] and [13] report results on other datasets, thus the comparison is not entirely relevant. ...
... Approach Performance Our approach 1 Statistical analysis [6] [5] 89% 95% ± 0.015 2 Statistical learning [14] 84% -3 ...
Deciding whether a handwritten signature is legit or it has been falsified is a very complex task. Several methods have been tried out by the graphology experts in order to detect such fraud. However, it is obvious that it is very hard to perform such a classification. In this paper we investigate the possibility to use some supervised learning techniques in order to build models capable to accurately perform such an analysis. The results reported during the testing phase of the obtained model are encouraging for further work.
... Signature verification and signature recognition are machine learning problems which try to learn from signature samples in order to decides and predict about the incoming signatures. Learning in an automatic signature verification/recognition, may be writer-independent (WI, general learning) or Writer-dependent (WD, special learning) [69]. In the first case, WI, learning is based on a large population of signature samples related to all persons in the dataset, whereas in the second case learning is based on the signature samples of each person, separately [69]. ...
... Learning in an automatic signature verification/recognition, may be writer-independent (WI, general learning) or Writer-dependent (WD, special learning) [69]. In the first case, WI, learning is based on a large population of signature samples related to all persons in the dataset, whereas in the second case learning is based on the signature samples of each person, separately [69]. Although WD learning achieve good results, for each user added to the system a classifier must be conducted which increases the complexity and cost of the system [18]. ...
... Along with the matching techniques, attention has been given to knowledge-base development also in relation to learning strategies [308], [310], [311] and signature modeling techniques [248], [308]. In particular, special attention has been given to writer-dependent learning strategies using only genuine specimens [156], [215], [216], [217], [328]. ...
... In other cases, the total error rate ε t , which is defined as ε t = ((FRR · P (ω 1 )) + (FAR · P (ω 2 ))-where P (ω 1 ) and P (ω 2 ) are the a priori probabilities of classes of genuine signatures (ω 1 ) and forgeries (ω 2 ), is used [281]- [283]. The receiver operating characteristic (ROC) curve analysis is also applied to FRR versus FAR evaluation since it shows the ability of a system to discriminate genuine signatures from forged ones [see Fig. 10(b)] [309], [311]. ...
In recent years, along with the extraordinary diffusion of the Internet and a growing need for personal verification in many daily applications, automatic signature verification is being considered with renewed interest. This paper presents the state of the art in automatic signature verification. It addresses the most valuable results obtained so far and highlights the most profitable directions of research to date. It includes a comprehensive bibliography of more than 300 selected references as an aid for researchers working in the field.
... The CEDAR Signature dataset [26] consists of 2640 signatures comprising 24 genuine signatures and 24 forged signatures for each user. It involves a total of 55 individuals with each person providing 48 signatures. ...
Deep learning is actively being used in biometrics to develop efficient identification and verification systems. Handwritten signatures are a common subset of biometric data for authentication purposes. Generative adversarial networks (GANs) learn from original and forged signatures to generate forged signatures. While most GAN techniques create a strong signature verifier, which is the discriminator, there is a need to focus more on the quality of forgeries generated by the generator model. This work focuses on creating a generator that produces forged samples that achieve a benchmark in spoofing signature verification systems. We use CycleGANs infused with Inception model-like blocks with attention heads as the generator and a variation of the SigCNN model as the base Discriminator. We train our model with a new technique that results in 80% to 100% success in signature spoofing. Additionally, we create a custom evaluation technique to act as a goodness measure of the generated forgeries. Our work advocates generator-focused GAN architectures for spoofing data quality that aid in a better understanding of biometric data generation and evaluation.
... However, it is reported that there will be some variations in the accuracy results due to the small number of samples that have been utilized in the training phase when applying this approach to verify signatures in the offline state. In terms of signature verification, two machine learning methods were sequentially presented in [22], and [23], genuine and forgery sets were involved in the general set, and the other involved only the genuine set. In the first method, counterexamples with near misses have been utilized in the learning process. ...
One of the most prevalent behavioral biometrics is the signature. In this paper, signatures are utilized in the case of recognition. Multiple contributions are provided here. Firstly, statistical analysis of efficiency is taken into consideration for the feature extraction. Secondly, a novel classifier is suggested. It is employed to recognize the signatures and it is called the Normalized Generalization Neural Network (NGNN). In terms of error rates, comparisons are established between different neural networks in the literature and the novel NGNN. The proposed NGNN consists of the input layer, normalization layer, Radial Basis Function (RBF) layer, and output layer. It can be considered as an enhanced or developed version of the Generalized Regression Neural Network (GRNN). A large number of signatures' attributes from the Biometric Ideal Test (BIT) database is utilized. That is, 1750 patterns of attributes are exploited. A significant improvement in the error rates over previous networks is achieved when using the novel NGNN. The Mean Absolute Error (MAE) has reached 0.028 and the Mean Square Error (MSE) has obtained 0.014. In addition, further experimental results on the BIT database showed better Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) of 0.002 and 0.119, respectively.
... III. DATASET In this paper, we utilized the CEDAR dataset [22] comprising genuine and forged handwritten signatures of 55 individuals. Each person contributed a set of 24 genuine signatures, which they made themselves, and a set of 24 forged signatures created by someone else. ...
Handwritten signature verification is a crucial task
with applications spanning authentication, financial transactions,
and legal documents. In scenarios where only a single reference
signature is available, the challenge of accurate verification
becomes pronounced due to variations in writing styles, distortions, and limited labeled data. In this paper, we propose
a novel Siamese-Transformer network tailored for handwritten
signature verification using few-shot learning. By synergizing
Siamese neural networks and Transformer architectures, our
model excels in capturing contextual relationships and discerning
genuine from forged signatures. A triplet loss function facilitates
discriminative feature learning. Convolution layers extract local
features from an image, while the transformer component utilizes
these local features to capture global dependencies within signatures. Experimental results on benchmark datasets showcase the
model’s superior performance in few-shot verification scenarios,
marking it as a promising advancement in signature verification
and few-shot learning techniques
... Many systems developed to combat a forgery depend on the fact someone's signature is unique according to the pattern of strokes he/she performs. In order to create s successful signature verification system at least these two steps is required: 1) Extracting the signature feature, and 2) Measure the difference between the two signatures in question [7]. CNN or Convolutional Neural Network has been a primary tool for tackling machine learning problems related to computer vision [5]. ...
The use of signatures is often encountered in various public documents ranging from academic documents to business documents that are a sign that the existence of signatures is crucial in various administrative processes. The frequent use of signatures does not mean a procedure without loopholes, but we must remain vigilant against signature falsification carried out with various motives behind it. Therefore, in this study, a signature verification system was developed that could prevent the falsification of signatures in public documents by using digital imagery of existing signatures. This study used neural networks with siamese network-based architectures that also empower self-supervised learning techniques to improve accuracy in the realm of limited data. The final evaluation of the machine learning method used gets a maximum accuracy of 83% and this result is better than the machine learning model that does not involve self-supervised learning methods.
... The data used in this study is the CEDAR Signature Verification Dataset 2 reported in [44]. In the study, 55 individuals contributed 24 of their own signatures, and some of the participants forged the signatures of others. ...
Forgery of a signature with the aim of deception is a serious crime. Machine learning is often employed to detect real and forged signatures. In this study, we present results which argue that robotic arms and generative models can overcome these systems and mount false-acceptance attacks. Convolutional neural networks and data augmentation strategies are tuned, producing a model of 87.12% accuracy for the verification of 2,640 human signatures. Two approaches are used to successfully attack the model with false-acceptance of forgeries. Robotic arms (Line-us and iDraw) physically copy real signatures on paper, and a conditional Generative Adversarial Network (GAN) is trained to generate signatures based on the binary class of 'genuine' and 'forged'. The 87.12% error margin is overcome by all approaches; prevalence of successful attacks is 32% for iDraw 2.0, 24% for Line-us, and 40% for the GAN. Fine-tuning with examples show that false-acceptance is preventable. We find attack success reduced by 24% for iDraw, 12% for Line-us, and 36% for the GAN. Results show exclusive behaviours between human and robotic forgers, suggesting training wholly on human forgeries can be attacked by robots, thus we argue in favour of fine-tuning systems with robotic forgeries to reduce their prevalence.
... Machine Learning algorithms learn the hidden pattern from the data is widely being used in different research problems [4] [5] [6] [4]. Similarly, Machine learning (ML) algorithms are being used in developing intelligent signature fraud detection system for offline signature verification, while image processing techniques are used to process the image of the signature on the paper for the implementation of the ML algorithms [7] [8]. Unlike traditional approaches of signature fraud detection, such as manual signature verification by human involvement, ML-based detection can identify previously undetected signature frauds and can deliver improved detection efficiency and efficacy [2]. ...
Signature fraud around the world is increasing atan alarming rate. Fraud in the signature may harm a personor an organization by false transactions and false documentauthorization, which may lead to an irreversible loss. Thus, thepurpose of this research is to predict forged signature usingmachine learning techniques. To attain the objective, differentstate-of-the-art machine learning models, including Neural Net-work, K-Nearest Neighbors, Support Vector Machine, DecisionTree, and Random Forest Classifier, were developed to classifybetween fraud and real signatures. To enhance the performanceof the Neural Network, the VGG-16 pre-trained model was used.As outcome, the transfer learning based Neural Network modelshowed the highest accuracy-96.7%, followed by Support VectorMachine (81.7%), K-Nearest Neighbors (71.7%), Random Forest(70.0%), and Decision Tree (68.3%).
(PDF) An Efficient Transfer Learning Model for Predicting Forged (Handwritten) Signature. Available from: https://www.researchgate.net/publication/360507780_An_Efficient_Transfer_Learning_Model_for_Predicting_Forged_Handwritten_Signature#fullTextFileContent [accessed Jan 28 2024].
... The data used in this study is the CEDAR Signature Verification Dataset 2 from [27]. In the study, 55 individuals contributed 24 of their own signatures, and some of the participants forged the signatures of others. ...
This study explores how robots and generative approaches can be used to mount successful false-acceptance adversarial attacks on signature verification systems. Initially, a convolutional neural network topology and data augmentation strategy are explored and tuned, producing an 87.12% accurate model for the verification of 2,640 human signatures. Two robots are then tasked with forging 50 signatures, where 25 are used for the verification attack, and the remaining 25 are used for tuning of the model to defend against them. Adversarial attacks on the system show that there exists an information security risk; the Line-us robotic arm can fool the system 24% of the time and the iDraw 2.0 robot 32% of the time. A conditional GAN finds similar success, with around 30% forged signatures misclassified as genuine. Following fine-tune transfer learning of robotic and generative data, adversarial attacks are reduced below the model threshold by both robots and the GAN. It is observed that tuning the model reduces the risk of attack by robots to 8% and 12%, and that conditional generative adversarial attacks can be reduced to 4% when 25 images are presented and 5% when 1000 images are presented.
... In this work on Machine Learning for Signature Verification [8] the author proposed implementation of how simple models could be used to identify and distinguish between forged and real signatures. Because each individual develops distinctive pen motion practices that serve to depict his or her signature, signatures are used for identification. ...
Signatures are widely used to validate the
authentication of an individual. A robust method is still
awaited that can correctly certify the authenticity of a
signature. The proposed solution provided in this paper is
going to help individuals to distinguish signatures for
determining whether a signature is forged or genuine. In our
system, we aimed to automate the process of signature
verification using Convolutional Neural Networks. Our model
is constructed on top of a pre-trained Convolutional Neural
Network called the VGG-19. We evaluated our model on
widely accredited signature datasets with a multitude of
genuine signature samples sourced from ICDAR[3],
CEDAR[1] and Kaggle[2]; achieving accuracies of 100%, 88%,
and 94.44% respectively. Our analysis shows that our
proposed model can classify the signature if they do not closely
resemble the genuine signature.
... An optimal size of database is also a very important factor in performing accurate biometric signature authentication. Another very interesting extension would be to see effect of applying a combination of machine learning (Srinivasan et al., 2006) and soft computing algorithms, e.g., fuzzy-neuro-genetic hybrid algorithm as very few have attempted it in the past. It uses genetic algorithm (GA) for the evolution of weight matrices of the fuzzified multilayer feed-forward back propagation neural network (BPNN). ...
In recent years, biometric signature verification (BSV) has been considered with renewed interest with increasing need of security and individual verification and authentication whether in banks, offices, institutions or other commercial organisations. Biometric signature verification is a behavioural biometric technique as a signature signifies unique behaviour of an individual. It can upgrade online banking using online digital systems for signing which cannot be altered or manipulated. Digital signature pads use algorithms to record the features of the signature, which is used to authenticate a signer during a transaction. This paper aims to present a comprehensive literature survey of the most recent research papers on biometric signature verification. It highlights the most important methods and addresses variations in the methods and features that are being taken up in the most recent research in this field along with the possible extensions.
Writer-independent offline signature verification is a widely utilized biometric authentication method and an area of active research. However, Writer-independent offline signature verification poses unique challenges in the Arabic language context. Despite the availability of benchmarking datasets such as CEDAR, and BHSig260 for Writer-independent offline signature verification research for languages such as English, Hindi, and Bengali, there is a notable absence of the availability of Arabic datasets. In this paper, we address this gap by introducing an Arabic signature verification dataset that serves as a benchmark for cross-lingual transfer learning evaluation as a cross-validation dataset. To establish a baseline for cross-lingual transfer learning for Arabic writer-independent offline signature verification, we adopt the Siamese neural network architecture. The results indicate that the dataset is comparably performing to existing benchmarks on cross-language transfer learning scenarios setting a solid baseline for the research potential in cross-lingual transfer learning. Moreover, we explore a voting ensemble approach that further enhances the accuracy of the model, highlighting the potential for continued research and improvement in this domain.
Forensic questioned document examiners still largely rely on visual assessments and expert judgment to determine the provenance of a handwritten document. Here, we propose a novel approach to objectively compare two handwritten documents using a deep learning algorithm. First, we implement a bootstrapping technique to segment document data into smaller units, as a means to enhance the efficiency of the deep learning process. Next, we use a transfer learning algorithm to systematically extract document features. The unique characteristics of the document data are then represented as latent vectors. Finally, the similarity between two handwritten documents is quantified via the cosine similarity between the two latent vectors. We illustrate the use of the proposed method by implementing it on a variety of collections of handwritten documents with different attributes, and show that in most cases, we can accurately classify pairs of documents into same or different author categories.
A person’s handwritten signature, one of the methods frequently used to confirm their identity, is used exclusively to confirm the biometric identification of different financial, legal, banking, insurance, and other business documents. Signature recognition techniques are used to identify which user someone’s signature is affiliated with. In recent years, many researchers have worked on applying new methods to this work, and deep learning methods have become quite prevalent among them. In order to provide more researchers with a better comprehension of how offline handwritten signature recognition work has evolved, the existing approaches, different architectures, challenging issues, and trends within the last 15 years, this paper follows a protocol to organize this work, collects information primarily from studies published in four major databases, applies inclusion and exclusion criteria, reviews offline handwritten signature recognition methods, including issues such as feature extraction and classification, and attempts to summarize the challenges and opportunities in the field. This paper emphasizes the popularity of research directions in this research area in deep learning. In contrast to other surveys in the field, this paper is not limited to a particular phase of work but provides a detailed account of each stage with the expectation that this will help future researchers.
The signature of the customer is still verified manually by the bank sectors. This includes new account creation, existing account merging, withdrawal form signature, cheque signature and many such. Which may lead to the rise in risk on the mutual basis? Many banks include both public sector banks and non-public sector banks. Where still human handles the work of signature matching and signs for the approval. This will lead to lot of corruptions; signature miss matching issues and many such issues. This is all because of the term called as “Human Error”. Human errors are unavoidable and not preventable; they are different from the machine errors. Based on these constraints in this document, a feasible technique is been planned. The solution to avoids the human error is banks has to choosing the computer recognizers for the signature verification, which is been built by the Robotic Process Automation. This process helps in verifying the customer’s signature accurately, this method is based on RPA which consist of the machine learning which is been deployed on the background of the RPA Orchestrator. The result is obtained by comparing both the signature which is signed during the bank joining application (Stored on bank Database). If signature doesn’t match it won’t move to the next process and the accuracy level is indicated. It is very important to keep records; hence a complete record is created with the details of customers name, accuracy level and matching (True/False). If true, it fetches the account details and process the withdraw form with a verified stamp.
Nowadays, the verification of handwritten signatures has become an effective research field in computer vision as well as machine learning. Signature verification is naturally formulated as a machine-learning task. This task is performed by determining if the signature is genuine or forged. Therefore, it is considered a two‐class classification issue. Since handwritten signatures are widely used in legal documents and financial transactions, it is important for researchers to select an efficient machine-learning technique for verifying these signatures and to avoid forgeries that may cause many losses to customers. So far, great outcomes have been obtained when using machine learning techniques in terms of equal error rates and calculations. This paper presents a comprehensive review of the latest studies and results in the last 10 years in the field of online and offline handwritten signature verification. More than 20 research papers were used to make a comparison between datasets, feature extraction, and classification techniques used in each system, taking into consideration the problems that occur in each. In addition, the general limitations and advantages of machine-learning techniques that are used to classify or extract signature features were summarized in the form of a table. We also present the general steps of the verification system and a list of the most considerable datasets available in online and offline fields.
One of the challenging and effective way of identifying person through biometric techniques is Signature verification as compared to traditional handcrafted system, where a forger has access and also attempt to imitate it which is used in commercial scenarios, like bank check payment, business organizations, educational institutions, government sectors, health care industry etc. so the signature verification process is used for human examination of a single known sample. There are mainly two types of signature verification: static and dynamic. i) Static or off-line verification is the process of verifying an electronic or document signature after it has been made, ii) Dynamic or on-line verification takes place as a person creates his/her signature on a digital tablet or a similar device. As compared, Offline signature verification is not efficient and slow for a large number of documents. Therefore although vast and extensive research on signature verification there is need to more focus and review on the online signature verification method to increase efficiency using deep learning.
Signature Verification is an essential and challenging task in forensic document analysis. Signature Verification is a process where the authenticity of a given signature is validated; i.e. it is confirmed, after analysis, whether the given signature is forged or not. The objective of the work proposed here is to create a new algorithm that can be used for verification of signatures used in cheques and various legal documents for customer authentication in banking systems. The proposed algorithm is designed using Daubechies complex wavelet transform (DCxWT) which has superior approximate shift invariant and edge representation properties over real valued wavelet transform. This is what motivates the use of DCxWT in the Signature Verification task. For classification of valid and forged signatures we have used Support Vector Machine (SVM) classifier. We have used Mutual Information kernel that is an alternative kernel of the KL divergence kernel in SVM classifier. Proposed method is compared with KL-divergence based kernel and some other standard kernels like Poly, RBF, and Tanh kernel of SVM Classifier. The method we proposed is juxtaposed with different state-of-the-art methods and effectiveness of the algorithm is also measured using different quantitative performance measures.
Cyber security comprises of technologies, architecture, infrastructure, and software applications that are designed to protect computational resources against cyber-attacks. Cyber security concentrates on four main areas such as application security, disaster security, information security, and network security. Numerous cyber security algorithms and computational methods are introduced by researchers to protect cyberspace from undesirable invaders and susceptibilities. But, the performance of traditional cyber security algorithms suffers due to different types of offensive actions that target computer infrastructures, architectures and computer networks. The implementation of intelligent algorithms in encountering the wide range of cyber security problems is surveyed, namely, nature-inspired computing (NIC) paradigms, machine learning algorithms, and deep learning algorithms, based on exploratory analyses to identify the advantages of employing in enhancing cyber security techniques.
We present in this paper a state of the latest advances in the field of offline handwritten signature verification. We describe the main approaches that have been proposed in the recent decades. Besides, we introduce the database of static signatures published in the literature as well as international competitions organized in the domain. Also, we present our contribution in the field.
This paper is a follow up to an article published in 1989 by R. Plamondon and G. Lorette on the state of the art in automatic signature verification and writer identification. It summarizes the activity from year 1989 to 1993 in automatic signature verification. For this purpose, we report on the different projects dealing with dynamic, static and neural network approaches. In each section, a brief description of the major investigations is given.
Handwriting has continued to persist as a means of communication and recording information in day-to-day life even with the introduction of new technologies. Given its ubiquity in human transactions, machine recognition of handwriting has practical significance, as in reading handwritten notes in a PDA, in postal addresses on envelopes, in amounts in bank checks, in handwritten fields in forms, etc. This overview describes the nature of handwritten language, how it is transduced into electronic data, and the basic concepts behind written language recognition algorithms. Both the online case (which pertains to the availability of trajectory data during writing) and the off-line case (which pertains to scanned images) are considered. Algorithms for preprocessing, character and word recognition, and performance with practical systems are indicated. Other fields of application, like signature verification, writer authentification, handwriting learning tools are also considered
We developed a system that automatically authenticates offline handwritten signatures using the discrete Radon transform (DRT) and a hidden Markov model (HMM). Given the robustness of our algorithm and the fact that only global features are considered, satisfactory results are obtained. Using a database of 924 signatures from 22 writers, our system achieves an equal error rate (EER) of 18% when only high-quality forgeries (skilled forgeries) are considered and an EER of 4.5% in the case of only casual forgeries. These signatures were originally captured offline. Using another database of 4800 signatures from 51 writers, our system achieves an EER of 12.2% when only skilled forgeries are considered. These signatures were originally captured online and then digitally converted into static signature images. These results compare well with the results of other algorithms that consider only global features.
This paper presents a set of geometric signature features for offline automatic signature verification based on the description of the signature envelope and the interior stroke distribution in polar and Cartesian coordinates. The features have been calculated using 16 bits fixed-point arithmetic and tested with different classifiers, such as hidden Markov models, support vector machines, and Euclidean distance classifier. The experiments have shown promising results in the task of discriminating random and simple forgeries.
In recent years, along with the extraordinary diffusion of the Internet and a growing need for personal verification in many daily applications, automatic signature verification is being considered with renewed interest. This paper presents the state of the art in automatic signature verification. It addresses the most valuable results obtained so far and highlights the most profitable directions of research to date. It includes a comprehensive bibliography of more than 300 selected references as an aid for researchers working in the field.
The decomposition of deformations by principal warps is demonstrated. The method is extended to deal with curving edges between landmarks. This formulation is related to other applications of splines current in computer vision. How they might aid in the extraction of features for analysis, comparison, and diagnosis of biological and medical images is indicated.
Introduction: The World of Documents The History of Questioned Document Examination -In Brief A Handwriting Compendium for Document Examiners The Discrimination of Handwriting The Premises for the Identification of Handwriting The Fundamentals of the Identification Process The Discrimination and Identification of Writing Special Problems in the Discrimination and Identification of Writing The Extrinsical Factors Influencing Handwriting The Intrinsical Influences upon Handwriting The Requirements and the Results The Diagnosis of Writing Identification The Scope of Document Examination The Sources of Document Examiners Science, Scientific Method and Writing Identification Graphology Understanding the Terms-Glossary Epilogue
New machine learning strategies are proposed for person identification which can be used in several biometric modalities such as friction ridges, handwriting, signatures and speech. The biometric or forensic performance task answers the question of whether or not a sample belongs to a known person. Two different learning paradigms are discussed: person-independent (or general learning) and person-dependent (or person-specific learning). In the first paradigm, learning is from a general population of ensemble of pairs, each of which is labelled as being from the same person or from different persons– the learning process determines the range of variations for given persons and between different persons. In the second paradigm the identity of a person is learnt when presented with multiple known samples of that person– where the variation and similarities within a particular person are learnt. The person-specific learning strategy is seen to perform better than general learning (5% higher perfor-mace with signatures). Improvement of person-specific performance with increasing number of samples is also observed.
This study is to examine the metric or non-metric prop-erties of binary vector dissimilarity measures, on which no comprehensive research has been conducted so far. Es-pecially, the triangle-inequality invariance characterizing several dissimilarity measures is revealed. Moreover, an entropy-based measurement is proposed to estimate the dis-criminative capability of a binary vector dissimilarity mea-sure, the effectiveness of the measurement is experimentally justified.
A graph matching approach to off-line signature verification is presented. Each of the signature images being compared is represented as a point set, which includes the local extremas of various types along the signature contours. Graph matching involves a deformation measure and a mapping function between point sets. Multiresolution signature features, as computed by gradient, structural and concavity features are used to measure local correspondence. Deformation and similarity scores are combined to make the final decision. The method performs better than previous off-line methods and is comparable to that of on-line systems
In this paper, a wavelet-based off-line handwritten signature verification system is proposed. The proposed system can automatically identify useful and common features which consistently exist within different signatures of the same person and, based on these features, verify whether a signature is a forgery or not. The system starts with a closed-contour tracing algorithm. The curvature data of the traced closed contours are decomposed into multiresolutional signals using wavelet transforms. Then the zero-crossings corresponding to the curvature data are extracted as features for matching. Moreover, a statistical measurement is devised to decide systematically which closed contours and their associated frequency data of a writer are most stable and discriminating. Based on these data, the optimal threshold value which controls the accuracy of the feature extraction process is calculated. The proposed approach can be applied to both on-line and off-line signature verification systems. Experimental results show that the average success rates for English signatures and Chinese signatures are 92.57% and 93.68%, respectively.
We describe novel methods of feature extraction for recognition of single isolated character images. Our approach is flexible in that the same algorithms can be used, without modification, for feature extraction in a variety of OCR problems. These include handwritten, machine-print, grayscale, binary and low-resolution character recognition. We use the gradient representation as the basis for extraction of low-level, structural and stroke-type features. These algorithms require a few simple arithmetic operations per image pixel which makes them suitable for real-time applications. A description of the algorithms and experiments with several data sets are presented in this paper. Experimental results using artificial neural networks are presented. Our results demonstrate high performance of these features when tested on data sets distinct from the training data.
There are inevitable variations in the signature patterns written by the same person. The variations can occur in the shape or in the relative positions of the characteristic features. In this paper, two methods are proposed to track the variations. Given the set of training signature samples, the first method measures the positional variations of the one-dimensional projection profiles of the signature patterns; and the second method determines the variations in relative stroke positions in the two-dimension signature patterns. The statistics on these variations are determined from the training set. Given a signature to be verified, the positional displacements are determined and the authenticity is decided based on the statistics of the training samples. For the purpose of comparison, two existing methods proposed by other researchers were implemented and tested on the same database. Furthermore, two volunteers were recruited to perform the same verification task. Results show that the proposed system compares favorably with other methods and outperforms the volunteers.
Existing word image retrieval algorithms suffer from either low retrieval precision or high computation complexity. We present an effective and efficient approach for word image matching by using gradient-based binary features. Experiments over a large database of handwritten word images show that the proposed approach consistently outperforms the existing best handwritten word image retrieval algorithm. Dynamic Time Warping (DTW) with profile-based shape features. Not only does the proposed approach have much higher retrieval accuracy, but also it is 893 times faster than DTW.
This paper describes a novel approach for signature verication and identication in an oine environment based on a quasi-multiresolution technique using GSC (Gradient, Structural and Concavity) features for feature extraction. These features when used at the word level, instead of the character level, yield promising results with accuracies as high as 78% and 93% for verication and identication, respectively. This method was successfully employed in our previous theory of individuality of handwriting devel- oped at CEDAR | based on obtaining within and between writer statistical distance distributions. In this paper, exploring signature verication and identication as oine handwriting verication and identication tasks respectively, we depict a mapping from the handwriting domain to the signature domain.
Massachusetts Institute of Technology. Dept. of Electrical Engineering. Thesis. 1970. Ph.D. MICROFICHE COPY ALSO AVAILABLE IN BARKER ENGINEERING LIBRARY. Vita. Bibliography: leaf 243. Ph.D.
The research here described centers on how a machine can recognize concepts and learn concepts to be recognized. Explanations are found in computer programs that build and manipulate abstract descriptions of scenes such as those children construct from toy blocks. One program uses sample scenes to create models of simple configurations like the three-brick arch. Another uses the resulting models in making identifications. Throughout emphasis is given to the importance of using good descriptions when exploring how machines can come to perceive and understand the visual environment.
In this paper we describe an algorithm that operates on the distances between features in the two related images and delivers a set of correspondences between them. The algorithm maximizes the inner product of two matrices, one of which is the desired 'pairing matrix' and the other a 'proximity matrix' with elements exp (-rij2/2 sigma 2), where rij is the distance between two features, one in each image, and sigma is an adjustable scale parameter. The output of the algorithm may be compared with the movements that people perceive when viewing two images in quick succession, and it is found that an increase in sigma affects the computed correspondences in much the same way as an increase in interstimulus interval alters the perceived displacements. Provided that sigma is not too small the algorithm will recover the feature mappings that result from image translation, expansion or shear deformation--transformations of common occurrence in image sequences--even when the displacements of individual features depart slightly from the general trend.
Motivated by several rulings in United States courts concerning expert testimony in general, and handwriting testimony in particular, we undertook a study to objectively validate the hypothesis that handwriting is individual. Handwriting samples of 1,500 individuals, representative of the U.S. population with respect to gender, age, ethnic groups, etc., were obtained. Analyzing differences in handwriting was done by using computer algorithms for extracting features from scanned images of handwriting. Attributes characteristic of the handwriting were obtained, e.g., line separation, slant, character shapes, etc. These attributes, which are a subset of attributes used by forensic document examiners (FDEs), were used to quantitatively establish individuality by using machine learning approaches. Using global attributes of handwriting and very few characters in the writing, the ability to determine the writer with a high degree of confidence was established. The work is a step towards providing scientific support for admitting handwriting evidence in court. The mathematical approach and the resulting software also have the promise of aiding the FDE.
Learning strategies and classification methods for verification of signatures from scanned documents are proposed and evaluated. Learning strategies considered are writer independent- those that learn from a set of signature sample (including forgeries) prior to enrollment of a writer, and writer dependent- those that learn only from a newly enrolled individual. Classification methods considered include two distance based methods (one based on a threshold, which is the standard method of signature verification and biometrics, and the other based on a distance probability distribution), a Nave Bayes (NB) classifier based on pairs of feature bit values and a support vector machine (SVM). Two scenarios are considered for the writer dependent scenario: (i) without forgeries (one-class problem) and (ii) with forgery samples being available (two class problem). The features used to characterize a signature capture local geometry, stroke and topology information in the form of a binary vector. In the one-class scenario distance methods are superior while in the two-class SVM based method outperforms the other methods.
Progress on the problem of signature verification has advanced more rapidly in online applications than offline applications, in part because information which can easily be recorded in online environments, such as pen position and velocity, is lost in static offline data. In offline applications, valuable information which can be used to discriminate between genuine and forged signatures is embedded at the stroke level. We present an approach to segmenting strokes into stylistically meaningful segments and establish a local correspondence between a questioned signature and a reference signature to enable the analysis and comparison of stroke features. Questioned signatures which do not conform to the reference signature are identified as random forgeries. Most simple forgeries can also be identified, as they do not conform to the reference signature's invariant properties such as connections between letters. Since we have access to both local and global information, our approach also shows promise for extension to the identification of skilled forgeries
words) is the key for obtaining the discriminating elements of handwriting. While allographs usually inhabit in words and segregation of a word into allographs is more subjective than objective, especially for cursive writing, analysis of handwritten words is a natural and better option. In this study, a handwritten word image is characterized by gradient, structural, and concavity features, and individuality of handwritten words is experimented through writership identification and verification on over 12,000 word images extracted from 3000 handwriting samples of 1000 individuals in U.S.. Experimental results show that handwritten words are very effective in differentiating handwriting and carry more individuality than most handwritten characters (allographs).
This paper is a description of recent advances in off-line signature verification research performed at our laboratory. Related works pertain to structural interpretation of signature images, directional PDF used as a global shape factor, the Extended Shadow Code (ESC) and the fuzzy ESC, a cognitive approach based on the Fuzzy ARTMAP, and shape factors related to visual perception.
Forensic Signature Examination
- S A Slyter
Slyter, S.A.: Forensic Signature Examination. Charles C. Thomas Pub (1995
Questioned Documents
- A Osborn
Signature testing in CEDAR-FOX. The display shows five known signatures, the questioned signature and the probability of the questioned matching the genuines
- Fig
Fig. 15. Signature testing in CEDAR-FOX. The display shows five known signatures, the questioned signature and the
probability of the questioned matching the genuines.
Off-line signature verification and identification using distance statistics <b>18<
- M K Kalera
- B Zhang
- S N Srihari