George S. Eskander

Hybrid writer-independent–writer-dependent offline signature verification system

Standard signature verification (SV) systems are writer-dependent (WD), where a specific classifier is designed for each individual. It is inconvenient to ask a user to provide enough number of signature samples to design his WD classifier. In practice, very few samples are collected and inaccurate classifiers maybe produced. To overcome this, writer-independent (WI) systems are introduced. A global classifier is designed using a development database, prior to enrolling users to the system. For these systems, signature templates are needed for verification, and the template databases can be compromised. Moreover, state-of-the-art WI and WD systems provide enhanced accuracy through information fusion at either feature, score or decision levels, but they increase computational complexity. In this study, a hybrid WI-WD system is proposed, as a compromise of the two approaches. When a user is enrolled to the system, a WI classifier is used to verify his queries. During operation, user samples are collected and adapt the WI classifier to his signatures. Once adapted, the resulting WD classifier replaces the WI classifier for this user. Simulations on the Brazilian and the GPDS signature databases indicate that the proposed hybrid system provides comparative accuracy as complex WI and WD systems, while decreases the classification complexity.

A bio-cryptographic system based on offline signature images

In bio-cryptography, biometric traits are replacing traditional passwords for secure exchange of cryptographic keys. The Fuzzy Vault (FV) scheme has been successfully employed to design bio-cryptographic systems as it can absorb a wide range of variation in biometric traits. Despite the intensity of research on FV based on physiological traits like fingerprints, iris, and face, there is no conclusive research on behavioral traits such as offline handwritten signature images, that have high inter-personal similarity and intra-personal variability. In this paper, a FV system based on the offline signature images is proposed. A two-step boosting feature selection (BFS) technique is proposed for selecting a compact and discriminant user-specific feature representation from a large number of feature extractions. The first step seeks dimensionality reduction through learning a population-based representation, that discriminates between different users in the population. The second step filters this representation to produce a compact user-based representation that discriminates the specific user from the population. This last representation is used to generate the FV locking/unlocking points. Representation variability is modeled by employing the BFS in a dissimilarity representation space, and it is considered for matching the unlocking and locking points during FV decoding. Proof of concept simulations involving 72,000 signature matchings (corresponding to both genuine and forged query signatures from the Brazilian Signature Database) have shown FV recognition accuracy of about 97% and system entropy of about 45-bits.

Signature based Fuzzy Vaults with Boosted Feature Selection

Handwritten signatures are commonly employed in many financial and forensic processes, and secure offline signature verification systems (SV) are required to automate such processes. In this context, bio-cryptography systems based on the handwritten signatures may be considered for enhance security. This paper presents a bio-cryptography system that constructs Fuzzy Vaults (FVs) based on the offline signature images. Boosting Feature Selection is employed to select features while training weak classifiers of offline SV systems. The indexes of selected features correspond to the most stable and discriminant features from a users signature images, and are used to encode user-specific FVs. A password is employed as a second authentication measure, to further enhance system security. During authentication, a user provides both the signature and the password to decode the FV and decouple his private key. If the FV is correctly decoded, the user is authenticated by the verification system. The proposed FV implementation alleviates the security vulnerabilities of the classical SV systems like template security, repudiation, irrevocability, and bypassing the classification decision. Moreover, simulations performed on a real-world signature verification database (with random, simple, and skilled forgeries) indicate security guarantees against stolen authentication measures. While compromised signatures or passwords lead to complete fail (FAR = 100%) of the classical SV or password protected cryptography systems respectively, compromised signatures lead to FAR of 0.1%, and compromised passwords leads to FAR of 15% with the proposed system.

Adaptation of Writer-Independent Systems for Offline Signature Verification

Although writer-independent offline signature verification (WI-SV) systems may provide a high level of accuracy, they are not secure due to the need to store user templates for authentication. Moreover, state-of-the-art writer-dependent (WD) and writer-independent (WI) systems provide enhanced accuracy through information fusion at either feature, score or decision levels, but they increase computational complexity. In this paper, a method for adapting WI-SV systems to different users is proposed, leading to secure and compact WD-SV systems. Feature representations embedded within WI classifiers are extracted and tuned to each enrolled user while building a user-specific classifier. Simulation results on the Brazilian signature database indicate that the proposed method yields WD classifiers that provide the same level of accuracy as that of the baseline WI classifiers (AER of about 5.38), while reducing complexity by about 99.5%.

On the dissimilarity representation and prototype selection for signature-based bio-cryptographic systems

Robust bio-cryptographic schemes employ encoding methods where a short message is extracted from biometric samples to encode cryptographic keys. This approach implies design limitations: 1) the encoding message should be concise and discriminative, and 2) a dissimilarity threshold must provide a good compromise between false rejection and acceptance rates. In this paper, the dissimilarity representation approach is employed to tackle these limitations, with the offline signature images are employed as biometrics. The signature images are represented as vectors in a high dimensional feature space, and is projected on an intermediate space, where pairwise feature distances are computed. Boosting feature selection is employed to provide a compact space where intra-personal distances are minimized and the inter-personal distances are maximized. Finally, the resulting representation is projected on the dissimilarity space to select the most discriminative prototypes for encoding, and to optimize the dissimilarity threshold. Simulation results on the Brazilian signature DB show the viability of the proposed approach. Employing the dissimilarity representation approach increases the encoding message discriminative power (the area under the ROC curve grows by about 47%). Prototype selection with threshold optimization increases the decoding accuracy (the Average Error Rate AER grows by about 34%).

Generation of Duplicated Off-Line Signature Images for Verification Systems

Biometric researchers have historically seen signature duplication as a procedure relevant to improving the performance of automatic signature verifiers. Different approaches have been proposed to duplicate dynamic signatures based on the heuristic affine transformation, nonlinear distortion and the kinematic model of the motor system. The literature on static signature duplication is limited and as far as we know based on heuristic affine transforms and does not seem to consider the recent advances in human behavior modeling of neuroscience. This paper tries to fill this gap by proposing a cognitive inspired algorithm to duplicate off-line signatures. The algorithm is based on a set of nonlinear and linear transformations which simulate the human spatial cognitive map and motor system intra-personal variability during the signing process. The duplicator is evaluated by increasing artificially a training sequence and verifying that the performance of four state-of-the-art off-line signature classifiers using two publicly databases have been improved on average as if we had collected three more real signatures.

A Dissimilarity-Based Approach for Biometric Fuzzy VaultsApplication to Handwritten Signature Images

Bio-Cryptographic systems enforce authenticity of cryptogra-phic applications like data encryption and digital signatures. Instead of simple user passwords, biometrics, such as, fingerprint and handwritten signatures, are employed to access the cryptographic secret keys. The Fuzzy Vault scheme (FV) is massively employed to produce bio-cryptogra-phic systems, as it absorbs variability in biometric signals. However, the FV design problem is not well formulated in the literature, and different approaches are applied for the different biometric traits. In this paper, a generic FV design approach, that could be applied to different biometrics, is introduced. The FV decoding functionality is formulated as a simple classifier that operates in a dissimilarity representation space. A boosting feature selection (BFS) method is employed for optimizing this classifier. Application of the proposed approach to offline signature biometrics confirms its viability. Experimental results on the Brazilian signature database (that includes various forgeries) have shown FV recognition accuracy of 90% and system entropy of about 69-bits.

Improving signature-based biometric cryptosystems using Cascaded Signature Verification-Fuzzy Vault (SV-FV) approach

Biometric cryptosystems have been applied to secure secret keys for encryption and digital signatures by means of biometric traits, e.g., Fingerprint, face, etc., where the fuzzy vault (FV) mechanism has been extensively employed. Recently, the authors proposed a FV system based on the offline signature images, so that digitized documents can be secured with the embedded handwritten signatures. However, the FV design concerns mostly with alleviating biometric variability with less focusing on its power in discriminating forgeries. Accordingly, the decoding accuracy of implementations is below the level required in practical banking transactions. On the other hand, signature verification (SV) systems have shown higher accuracy in discriminating forgeries. In this paper, accuracy of signature-based biometric cryptosystems is enhanced by cascading SV and FV modules. Signature samples are first verified by the SV module. Then, only verified samples are processed by FV decoders for unlocking cryptographic keys. Hence, the upper limit of the false accept rate is determined by the more accurate SV module. Simulation results obtained with the Brazilian signature database indicate the viability of the proposed approach. Cascaded SV-FV system increases decoding accuracy by about 35% compared to the pure FV systems.

Offline signature-based fuzzy vault: A review and new results

An offline signature-based fuzzy vault (OSFV) is a bio-cryptographic implementation that uses handwritten signature images as biometrics instead of traditional passwords to secure private cryptographic keys. Having a reliable OSFV implementation is the first step towards automating financial and legal authentication processes, as it provides greater security of sensitive documents by means of the embedded handwritten signatures. The authors have recently proposed the first OSFV implementation, where a machine learning approach based on the dissimilarity representation concept is employed to select a reliable feature representation adapted for the fuzzy vault scheme. In this paper, some variants of this system are proposed for enhanced accuracy and security. In particular, a new method that adapts user key size is presented. Performance of proposed methods are compared using the Brazilian PUCPR and GPDS signature databases and results indicate that the key-size adaptation method achieves a good compromise between security and accuracy. As the average system entropy is increased from 45-bits to about 51-bits, the AER (average error rate) is decreased by about 21%.