Thian Song Ong

Integrating Palmprint and Fingerprint for Identity Verification

In this paper, we propose a multimodal biometrics system that combines fingerprint and palmprint features to overcome several limitations of unimodal biometrics—such as the inability to tolerate noise, distorted data and etc.—and thus able to improve the performance of biometrics for personal verification. The quality of fingerprint and palmprint images are first enhanced using a series of pre-processing techniques. Following, a bank of 2D Gabor filters is used to independently extract fingerprint and palmprint features, which are then concatenated into a single feature vector. We conclude that the proposed methodology has better performance and is more reliable compared to unimodal approaches using solely fingerprint or palmprint biometrics. This is supported by our experiments which are able to achieve equal error rate (EER) as low as 0.91% using the combined biometrics features.

Secure Minutiae-Based Fingerprint Templates Using Random Triangle Hashing

Due to privacy concern on the widespread use of biometric authentication systems, biometric template protection has gained great attention in the biometric research recently. It is a challenging task to design a biometric template protection scheme which is anonymous, revocable and noninvertible while maintaining acceptable performance. Many methods have been proposed to resolve this problem, and cancelable biometrics is one of them. In this paper, we propose a scheme coined as Random Triangle Hashing which follows the concept of cancelable biometrics in the fingerprint domain. In this method, re-alignment of fingerprints is not required as all the minutiae are translated into a pre-defined 2 dimensional space based on a reference minutia. After that, the proposed Random Triangle hashing method is used to enforce the one-way property (non-invertibility) of the biometric template. The proposed method is resistant to minor translation error and rotation distortion. Finally, the hash vectors are converted into bit-strings to be stored in the database. The proposed method is evaluated using the public database FVC2004 DB1. An EER of less than 1% is achieved by using the proposed method.

Generating revocable fingerprint template using minutiae pair representation

With the growing concern for secure and private information protection, several issues pertaining to the traditional biometric authentication system such as the reissuance of biometric template and template security have been raised. To resolve these limitations, many new concepts of revocable biometrics have been introduced. However, the design of a template protection scheme that fulfils all the security and privacy criteria, namely revocability, non-invertibility, and anonymity, is still a challenging task. In this paper, we present a secure fingerprint template protection scheme by using minutiae pair representation that extends the concept of revocable biometrics. The proposed method integrates template with its discriminative helper data to achieve acceptable equal error rate (EER) of near 0%. Apart from performance, the security and privacy issues such as revocability and non-invertibility are also discussed.

A multiple layer fusion approach on keystroke dynamics

In this paper, we present a novel keystroke dynamic recognition system by means of a novel two-layer fusion approach. First, we extract four types of keystroke latency as the feature from our dataset. The keystroke latency will be transformed into similarity scores via Gaussian Probability Density Function (GPD). We also propose a new technique, known as Direction Similarity Measure (DSM), which measures the absolute difference between two sets of latency. Last, four fusion approaches coupled with six fusion rules are applied to improve the final result by combining the scores that are produced by GPD and DSM. Best result with equal error rate of 1.401% is obtained with our two-layer fusion approach.

Generating revocable fingerprint template using polar grid based 3-tuple quantization technique

Recently, biometric template protection has received much attention from the research community due to the security and privacy concerns of biometric template. Although a number of biometric template protection methods have been reported, yet, it is still a challenging task to devise a biometric template protection method which satisfies all the biometric template protection criteria namely diversity, revocability, non-invertibility and performance. In this paper, a method is proposed to generate a revocable fingerprint template of bit-string from a set of minutiae points through polar grid based 3-tuple quantization (PGTQ) technique. Two merits of the proposed method are outlined, i.e., alignment-free and performance. FVC2002 DB1 and DB2 are used to evaluate the performance of the proposed method. Besides, the template protection criteria are also examined to ensure the security and privacy of the biometric template.

Multimodal biometrics based bit extraction method for template security

In this paper, we propose a secure and revocable biometric bit-string generation technique for template protection. The proposed method consists of random tiling and equal probable discretisation. Random tiling is a feature transformation method to derive random features from biometric data based on a user specific key. In the event of template is compromised, a refreshed biometric template can easily be issued by replacing the compromised key with a new user specific key. On the other hand, we propose a modified equal probable discretisation to partitions the uneven biometric data distribution into different equal probable segments rather than equal width segments. This guarantees each set of the codeword has the same likelihood of occurring and thus user privacy is strengthened as it becomes difficult for an adversary to correctly guess the codeword associated with each segment. The proposed method is evaluated using multimodal biometrics — the fusion of fingerprint and palmprint at feature level. Encouraging experimental results vindicate the feasibility of our approach.

Dynamic Handwritten Signature Verification Based on Statistical Quantization Mechanism

Online handwritten signature has been widely used for identity verification. However, it suffers from large intra-class variation problem as individual’s signature may deviate from time to time due to variations in signing position, signature size, writing surface, and other factors. In addition, signatures are easier to forge than other biometrics and this leads to random and skilled forgeries issues. In this paper, we propose a novel Statistical Quantization Mechanism (SQM) to suppress the intra-class variation in signature features and thus discriminate the difference between genuine signature and its forgery. Experimental results show the proposed method is feasible in practice.

Enhanced maximum curvature descriptors for finger vein verification

Maximum Curvature Method (MCM) is one of the promising methods for finger vein verification. MCM scans the curvature of the vein image profiles within a finger for feature extraction. However, the quality of the image can be poor due to variations in illumination and sensor conditions. Furthermore, traditional MCM matching of the vein pattern requires extensive processing time. To address these limitations, we propose an integrated Enhanced Maximum Curvature (EMC) method with Histogram of Oriented Gradient (HOG) descriptor for finger vein verification. Unlike MCM, EMC incorporates an enhancement mechanism to extract small vein delineation that is hardly visible in the extracted vein patterns. Next, HOG is applied instead of image binarization to convert a two-dimensional vein image into a one-dimensional feature vector for efficient matching. The HOG descriptor is able to characterize the local spatial representation of a finger vein by capturing the gradient information effectively. The proposed method is evaluated based on two datasets namely the PKU Finger Vein Database (V4) and SDUMLA-HMT finger vein database. Experiments show promising verification results with equal error rates as low as 0.33 % for DB1 and 0.14 % for DB2 respectively, when EMC+HOG+SVM is applied.

Improved Biohashing Method Based on Most Intensive Histogram Block Location

Biohashing is a promising cancellable biometrics method. However, it suffers from a problem known as ‘stolen token scenario’. The performance of the biometric system drops significantly if the Biohashing private token is stolen. To solve this problem, this paper proposes a new method termed as Most Intensive Histogram Block Location (MIBL) to extract additional information of the p-th best gradient magnitude. Experimental analysis shows that the proposed method is able to solve the stolen token problem with error equal rates as low as 1.46% and 7.27% when the stolen token scenario occurred for both FVC2002 DB1 and DB2 respectively.

Typing dynamics biometric authentication through fuzzy logic

The request to protect sensitive data and computer systems from malicious impostors while allowing ease of access for legitimate users is one of the greatest challenges in the computer security field. Traditionally, conventional ID and passwords authentication schemes are used for controlling access to computer systems. However, this approach has many inherent flaws such as password sharing and password guessing. To confront these problems, we propose a dynamic typing biometric authentication method for computer security system. As a userpsilas typing template cannot be shared, lost or forgotten. It can be reliable used to gain access to the computer system. In this paper, we examine the behavioral biometrics of typing dynamics which uses the manner and rhythm in which an individual types characters on a keyboard or keypad. Experiment shows that an equal error rate (EER) of 20% could be achieved using the proposed approach. This paper focuses on the application of fuzzy logic in typing dynamic and explores the advantages of using this method as compare to the statistical standard deviation.