Showing papers on "Signature recognition published in 2016"

A new algorithm for identity verification based on the analysis of a handwritten dynamic signature

Abstract: Graphical abstractDisplay Omitted HighlightsWe propose a new algorithm for the dynamic signature verification.The algorithm implements a new way of signatures division - hybrid partitioning.Hybrid partitions are associated with time and dynamics of the signing process.The algorithm compares test signatures to the reference ones in interpretable way.The algorithm works independently for each signer. Identity verification based on authenticity assessment of a handwritten signature is an important issue in biometrics. There are many effective methods for signature verification taking into account dynamics of a signing process. Methods based on partitioning take a very important place among them. In this paper we propose a new approach to signature partitioning. Its most important feature is the possibility of selecting and processing of hybrid partitions in order to increase a precision of the test signature analysis. Partitions are formed by a combination of vertical and horizontal sections of the signature. Vertical sections correspond to the initial, middle, and final time moments of the signing process. In turn, horizontal sections correspond to the signature areas associated with high and low pen velocity and high and low pen pressure on the surface of a graphics tablet. Our previous research on vertical and horizontal sections of the dynamic signature (created independently) led us to develop the algorithm presented in this paper. Selection of sections, among others, allows us to define the stability of the signing process in the partitions, promoting signature areas of greater stability (and vice versa). In the test of the proposed method two databases were used: public MCYT-100 and paid BioSecure.

Performance of an Off-Line Signature Verification Method Based on Texture Features on a Large Indic-Script Signature Dataset

Abstract: In this paper, a signature verification method based on texture features involving off-line signatures written in two different Indian scripts is proposed. Both Local Binary Patterns (LBP) and Uniform Local Binary Patterns (ULBP), as powerful texture feature extraction techniques, are used for characterizing off-line signatures. The Nearest Neighbour (NN) technique is considered as the similarity metric for signature verification in the proposed method. To evaluate the proposed verification approach, a large Bangla and Hindi off-line signature dataset (BHSig260) comprising 6240 (260×24) genuine signatures and 7800 (260×30) skilled forgeries was introduced and further used for experimentation. We further used the GPDS-100 signature dataset for a comparison. The experiments were conducted, and the verification accuracies were separately computed for the LBP and ULBP texture features. There were no remarkable changes in the results obtained applying the LBP and ULBP features for verification when the BHSig260 and GPDS-100 signature datasets were used for experimentation.

Gait recognition based on modified phase-only correlation

Abstract: Gait recognition is an emerging biometric technology aiming to identify people purely through the analysis of the way they walk. The technology has attracted interest as a method of identification because it is noncontact and does not require the subject’s cooperation. Clothing, carrying conditions and other intra-class variations, also referred to as “covariates,” affect the performance of gait recognition systems. This paper proposes a supervised feature extraction method, which is able to select relevant discriminative features for human recognition to mitigate the impact of covariates and hence improve the recognition performances. The proposed method is evaluated using the CASIA gait database (dataset B), and the experimental results suggest that our method yields 81.40 % of correct classification when compared against similar techniques which do not exceed 77.96 %.

Study on CNN in the recognition of emotion in audio and images

Abstract: In this paper, the performance of Convolution Neural Network (CNN) in image recognition and emotion recognition in speech will be compared and presented. Feature extraction and selection in pattern recognition is an important issue and have been frequently discussed. Moreover, two-dimensional signals such as image and voice are hard to be modelled well by traditional models like SVM. The ability of CNN to characterize two-dimensional signals is prominent. And CNN can adaptively extract feature to eliminate the dependence on human subjectivity or experience. It mimics the effect of local filtering in visual cortex cells to dig local correlation in natural dimensional space. In this work, for the problems of the image recognition and emotion recognition in speech, CNN and SVM which is used as baseline for comparison of the recognition effect. Different kernel functions in SVM have been experimented for image recognition with, the best accuracy is 94.17%. However, the accuracy of using CNN is 95.5% (7291 pictures for train and 2007 pictures for test) with less time consuming. In the emotion recognition of speech, the accuracy of CNN is 97.6% corresponds to 55.5% by baseline model (4000 utterances for training, 1500 for validation, 500 for test). The experimental results showed that CNN can effectively extract features and its modeling capability for two-dimensional signals is prominent.

Modeling Time Series Similarity with Siamese Recurrent Networks

Abstract: Traditional techniques for measuring similarities between time series are based on handcrafted similarity measures, whereas more recent learning-based approaches cannot exploit external supervision. We combine ideas from time-series modeling and metric learning, and study siamese recurrent networks (SRNs) that minimize a classification loss to learn a good similarity measure between time series. Specifically, our approach learns a vectorial representation for each time series in such a way that similar time series are modeled by similar representations, and dissimilar time series by dissimilar representations. Because it is a similarity prediction models, SRNs are particularly well-suited to challenging scenarios such as signature recognition, in which each person is a separate class and very few examples per class are available. We demonstrate the potential merits of SRNs in within-domain and out-of-domain classification experiments and in one-shot learning experiments on tasks such as signature, voice, and sign language recognition.

Gender recognition from mobile biometric data

Abstract: This paper investigates gender recognition from keystroke dynamics data and from touchscreen swipes. Classification measurements were performed using 10-fold cross-validation and leave-one-user-out cross-validation (LOUOCV). We show that when the target is unseen user data classification, only the second approach is viable. Based on our limited datasets, we show that gender cannot be reliably predicted. The best results were 64.76% for the keystroke dataset and 57.16% for the swipes dataset. However, the classification accuracy is over 80% for more than half of the users in the case of keystroke dynamics dataset.

Authentic mobile-biometric signature verification system

Abstract: This is an undeniable fact that in the coming years a considerable percentage of organisations are drifting toward mobile devices for authentication. Banking sector as an additional offshoot has shifted to mobile devices with their applications for e-banking and mobile-banking, giving rise to an emergent requirement of a foolproof and authentic mobile-biometric system. This study presents an authentic mobile-biometric signature verification system and a comparative analysis of the performance of the proposed system for the two datasets; one using the standard device that is used for capturing biometric signatures and the other one is a mobile database taken from a smart phone for biometric signature authentication. The results presented demonstrate that the proposed system outperforms existing mobile-biometric signature verification systems based on dynamic time warping and hidden Markov model. Moreover, this study presents a comprehensive survey of mobile-biometric systems, different devices and hardware needed to support mobile biometrics along with open issues and challenges faced by the mobile-biometric systems. The experiments presented establish that the performance of mobile devices is low as compared with normal biometric signature capturing devices and the major reason the authors found is the absence of pen-tilt angle information in the mobile device datasets.

Offline signature verification based on geometric feature extraction using artificial neural network

Abstract: Signature verification is widely used for personal verification. But, it has an inevitable problem of getting exploited for forgery therefore an automatic signature recognition and verification system is required. Verification can be accomplished either Online or Offline based application. Offline systems work on the scanned image of a signature. Online systems use dynamic information like pressure, speed etc. of a signature during the time when the signature is made. In this paper, we present an offline signature verification technique based on geometric features. We have used six geometric features namely Area, Centroid, Standard deviation, Even pixels, Kurtosis and Skewness. In our technique first the preprocessing of a scanned signature image is done to isolate the signature and to remove noise. The system is trained using a database of signatures obtained from authenticated users. Then artificial neural network (ANN) is used in recognition and verification of signatures: genuine or forged, and efficiency is about 89.24% having threshold of 80%. Simulation results shows that the technique is robust and clearly differentiates between genuine and forgery signatures.

Recognition of handwritten digits using the signature features and Optimum-Path Forest Classifier

Abstract: There is a growing need for recognition of digits manuscripts for use in various situations, such as recognition of handwritten postal address digits for automated redirection of letters in the mail, acknowledgment of nominal values in bank checks. Recognition of handwritten digits faces great difficulty in dealing with intra-class variation due to different writing styles, different degrees of inclination of the characters. Optical character recognition systems, also known as OCR, identifying and recognizing printed characters through images, an already widespread functionality in scanners, mobile devices, among others. This paper presents the use of the classifier Optimum-Path Forest (OPF) applied in handwriting recognition digits. A new feature extraction method is proposed using signature of the characters, and the OPF algorithm is used in the classification. According to the results presented, it appears that the detection and recognition of characters are being carried out satisfactorily in the Manhattan distance stood out with an average accuracy of 99.53%, and get training times and test lower than the other methods such as It is the characteristic of OPF method.

An offline signature verification technique using pixels intensity levels

Abstract: Offline signature recognition has great importance in our day to day activities. Researchers are trying to use them as biometric identification in various areas like banks, security systems and for other identification purposes. Fingerprints, iris, thumb impression and face detection based biometrics are successfully used for identification of individuals because of their static nature. However, people’s signatures show variability that makes it difficult to recognize the original signatures correctly and to use them as biometrics. The handwritten signatures have importance in banks for cheque, credit card processing, legal and financial transactions, and the signatures are the main target of fraudulence. To deal with complex signatures, there should be a robust signature verification method in places such as banks that can correctly classify the signatures into genuine or forgery to avoid financial frauds. This paper, presents a pixels intensity level based offline signature verification model for the correct classification of signatures. To achieve the target, three statistical classifiers; Decision Tree (J48), probability based Naive Bayes (NB tree) and Euclidean distance based k-Nearest Neighbor (IBk), are used. For comparison of the accuracy rates of offline signatures with online signatures, three classifiers were applied on online signature database and achieved a 99.90% accuracy rate with decision tree (J48), 99.82% with Naive Bayes Tree and 98.11% with K-Nearest Neighbor (with 10 fold cross validation). The results of offline signatures were 64.97% accuracy rate with decision tree (J48), 76.16% with Naive Bayes Tree and 91.91% with k-Nearest Neighbor (IBk) (without forgeries). The accuracy rate dropped with the inclusion of forgery signatures as, 55.63% accuracy rate with decision tree (J48), 67.02% with Naive Bayes Tree and 88.12% (with forgeries).

Face recognition based on matching of local features on 3D dynamic range sequences

Abstract: 3D face recognition has attracted attention in the last decade due to improvement of technology of 3D image acquisition and its wide range of applications such as access control, surveillance, human-computer interaction and biometric identification systems. Most research on 3D face recognition has focused on analysis of 3D still data. In this work, a new method for face recognition using dynamic 3D range sequences is proposed. Experimental results are presented and discussed using 3D sequences in the presence of pose variation. The performance of the proposed method is compared with that of conventional face recognition algorithms based on descriptors.

On the use of fingernail images as transient biometric identifiers

Abstract: The significant advantages that biometric recognition technologies offer are in danger of being left aside in everyday life due to concerns over the misuse of such data. The biometric data employed so far focuses on the permanence of the characteristics involved. A concept known as `the right to be forgotten' is gaining momentum in international law and this should further hamper the adoption of permanent biometric recognition technologies. However, a multitude of common applications are short-term and, therefore, non-permanent biometric characteristics would suffice for them. In this paper we discuss `transient biometrics,' i.e. recognition via biometric characteristics that will change in the short term and show that images of the fingernail plate can be used as a transient biometric with a useful life-span of less than 6 months. A direct approach is proposed that requires no training and a relevant evaluation dataset is made publicly available.

Face recognition based on subject dependent Hidden Markov Models

Abstract: In this paper we present an automatic face recognition system based on incremental Singular Values Decomposition (SVD) and subject dependent Hidden Markov Models (HMM). For each subject, an individual HMM is trained with features, extracted from the orthogonal decomposition (SVD) of the subject's training images. The main advantage of the proposed SVD-HMM recognition system is the robustness against image dimensionality reduction. The system was tested on two benchmark face datasets — the Olivetti Research Laboratory (ORL) and the YALE database. The SVD-HMM was further compared with a standard SVD face recognition. SVD applied to the original (full size) images performs similarly to the SVD-HMM applied to the compressed (half of the original size) images. SVD degrades rapidly when the image is compressed.

Signature recognition: establishing human baseline performance via crowdsourcing

Abstract: This work explores crowdsourcing for the establishment of human baseline performance on signature recognition. We present five experiments according to three different scenarios in which laymen, people without Forensic Document Examiner experience, have to decide about the authenticity of a given signature. The scenarios include single comparisons between one genuine sample and one unlabeled sample based on image, video or time sequences and comparisons with multiple training and test sets. The human performance obtained varies from 7% to 80% depending of the scenario and the results suggest the large potential of these collaborative platforms and encourage to further research on this area.

Stride Towards Proposing Multi-Modal Biometric Authentication for Online Exam

Abstract: Biometric authentication has been getting widespread attention over the past decade with growing demands in automated secured personal identification and has been employed in diverse fields. It ensures actual presence of biometric entity of a person in contrast to a fake self-manufactured synthetic or reconstructed sample is a significant problem. Also in the previous work they use face and dress color as hard and soft biometric traits. The major drawback of the existing continuous authentication system is, it is able to successfully authenticate the user continuously with high tolerance to the user posture. So, to overcome this drawback and improve the systems robustness against illumination changes and cluttered background, in this paper we use additional biometric traits which are mole, ornament details and face dimensions in addition to the dress color and face color. Also, we extend it to the online exam application. That is, continuously monitoring of a person in an online exam is proposed employing hard biometric like facial recognition and soft biometrics. Modified PCA (Principal Component Analysis) is employed here for the facial recognition part. Both the hard biometric (face) and soft biometrics is fused with the help of optimization algorithm based similarity technique. Finally the authentication is performed and evaluated using standard evaluation metrics. The technique is implemented in MATLAB and will be compared to prominent existing techniques.

Online Signature Verification Based on Biometric Features

Abstract: Since current signatures are generally not verified carefully, frauds by forging others signature always happen. This paper tried to authenticate user automatically with electronic signatures on mobile device. We collected coordinates, pressure, contact area and other biometric data when users sign their name on touch screen smart phone. Then we used four different classification algorithms, Support Vector Machine, Logistic Regression, AdaBoost and Random Forest to build a specific signature verification model for each user, and compared the verification accuracy of these algorithms. The experimental result on 42 persons' dataset shows that these four algorithms have satisfactory performance on Chinese signature verification, and Adaboost has the best performance with error rate of 2.375%.

Towards human-assisted signature recognition: Improving biometric systems through attribute-based recognition

Abstract: This work explores human-assisted schemes for improving automatic signature recognition systems. We present a crowdsourcing experiment to establish the human baseline performance for signature recognition tasks and a novel attribute-based semi-automatic signature verification system inspired in FDE analysis. We present different experiments over a public database and a self-developed tool for the manual annotation of signature attributes. The results demonstrate the benefits of attribute-based recognition approaches and encourage to further research in the capabilities of human intervention to improve the performance of automatic signature recognition systems.

Handwritten signature recognition method based on fuzzy logic

Abstract: We suggested a method for handwritten signature recognition based on fuzzy logic. First of all, we proposed some features of handwritten signature based on curvature properties with fuzzy values. Then we proposed a method for signature recognition based on comparing these fuzzy features. We used collection of signatures MCYT_Signature_100 for testing our method. Signature recognition experiment has been conducted with 100 users, 25 original and 25 fake signatures for each user. As a result, we have got FRR value 0.03 and FAR value 0.01 which are better than the results of some other methods.

On-line signature verification using point pattern matching algorithm

Abstract: Authentication of persons has been known as a paramount part in society. Security requirements have given biometrics much attention. Verification of the signature is one of the biometric methods used in recognition systems. This paper presents a new method to verify an online signature of a person. In features extraction step, local parameters are extracted as time functions of different dynamic properties. In recognition phase the Discrete to Continuous Algorithm is applied on features to verify the signature claimed by a person. The results obtained by the proposed algorithm show a good accuracy rate.

An efficient star skeleton extraction for human action recognition using hidden Markov models

Abstract: This paper aims at finding an efficient approach for automatic human action recognition to classify human actions in both outdoor and indoor environments. A human action recognition system (HARS) collects video frames of human activities, extracts the desired features of each human skeleton. These characteristics are calculated, classified to build a skeleton database that can distinguish almost human gestures. This HARS converts every sequence of human gestures to the sequences of skeletal joint mapping (SJM). Then it assigns corresponding observation symbols to each SJM. Those observation sequences are used to train of hidden Markov models (HMMs) corresponding to seven actions: standing, walking, running, jumping, falling, lying, and sitting. Baum-Welch and forward-backward algorithms are employed to find optimal parameters of each HMM. During a recognition phase, each human gesture sequence is converted to an observation sequence and put into seven optimized HMM models. The current action can be identified by finding a model with the highest probability. The experimental results show that the proposed HARS offers high accuracy of action recognition in real-time.

Empirical mode decomposition for online handwritten signature verification

Abstract: The handwritten signature is a biometric method used to verify a person's identity. This study lies within the scope of an online handwritten signature verification system, in which a signature is modelled by an analytical approach based on the empirical mode decomposition. The organised system is tested on the SVC2004 task1 and MYCT-100 databases. The implemented evaluation protocol shows the importance of the adopted method and allows obtaining an equal error rate of 1.83 and 2.23% for the SVC2004 task1 and the MYCT-100 databases, respectively.

Facial expression recognition based on static and dynamic approaches

Abstract: The identification of facial expressions with human emotions plays a key role in non-verbal human communication and has applications in several areas. In this work, we analyze two main approaches for expression recognition.

Simplified scoring methods for HMM-based speech recognition

Abstract: Most of the contemporary speech recognition systems exploit complex algorithms based on Hidden Markov Models (HMMs) to achieve high accuracy. However, in some cases rich computational resources are not available, and even isolated words recognition becomes challenging task. In this paper, we present two ways to simplify scoring in HMM-based speech recognition in order to reduce its computational complexity. We focus on core HMM procedure--forward algorithm, which is used to find the probability of generating observation sequence by given HMM, applying methods of dynamic programming. All proposed approaches were tested on Russian words recognition and the results were compared with those demonstrated by conventional forward algorithm.

Online human action recognition based on improved dynamic time warping

Abstract: Online human action recognition has broad application prospect in many fields of computer vision. Simultaneously, with the advent of depth camera, it brings on a new trend of online human action recognition but still present some unique challenges. In this paper, to solve the lower accuracy of the existing online human action recognition algorithm based on depth camera, we adopt the improved Dynamic Time Warping (DTW) algorithm to reduce the pathologic alignment caused by traditional DTW algorithm in features extraction and template generation, and then the recognition accuracy will be enhanced. Finally, this proposed approach will be evaluated on MSRC-12 Kinect Gesture dataset. The experimental evaluations show that the proposed approach achieves superior performance to the state of the art algorithms.

Multimodal biometric template authentication of finger vein and signature using visual cryptography

Abstract: In this paper personal verification method using finger-vein and signature is presented. Among many authentication systems finger-vein is promising as the foolproof method of automatic personal identification. Finger-vein and signature image is pre-processed and features are extracted using cross number concept and principle compound analysis. Fusion technique is used to fuse the finger vein and signature images. Then the visual cryptographic scheme is applied for the biometric template to generate the shares. The shares are stored in a separate database, and then the biometric image is revealed only when both the shares are simultaneously available. At the same time, the individual image does not reveal the identity of the biometric image. The proposed work is evaluated with evaluation metrics FAR, FRR and accuracy.

Design of multimodal biometrics system based on feature level fusion

Abstract: Multimodal system aims to fuse two or more biometrics traits of an individual to achieve improvement in FAR and FRR of biometrics system which in turn increases accuracy of system. In this paper we have proposed biometrics system based on biometrics traits face and signature. The performances of face and signature recognition can be enhanced using a proposed feature selection method to select an optimal subset of features. Signature is very important human characteristics which is required in all financial transaction for human identification. In case of financial transaction correct recognition is necessary otherwise it can lead to fraudulent activities. Face is most commonly acceptable and popular biometrics. Proposed algorithm fuses wavelet based features of face and signature. Wavelet based feature fusion method also gave very promising results. Hamming distance classifier is used to take decision whether person is genuine or imposter. Our experiments show that the proposed algorithm can achieve higher classification accuracy than offline signature and face based identification system. We have achieved false accept rate of 5.99% and 3% for multibiometrics system for ORL databases combined with Caltech and Ucoer real signature database resp.

Flexible Neural Network Architecture for Handwritten Signatures Recognition

Abstract: This article illustrates modeling of flexible neural networks for handwritten signatures preprocessing. An input signature is interpolated to adjust inclination angle, than descriptor vector is composed. This information is preprocessed in proposed flexible neural network architecture, in which some neurons are becoming crucial for recognition and adapt to classification purposes. Experimental research results are compared in benchmark tests with classic approach to discuss efficiency of proposed solution.