Amit Kaul

A survey of emerging biometric modalities

Biometric recognition is the task of identifying an individual on the basis of his/her physiological or behavioral traits. Necessity of developing fool proof security systems has provided biometric research much needed impetus. Over the last three decades there has been a lot of work done on development of systems based on fingerprint, face, iris, voice etc., but in the recent past some new biometric measures have emerged which have shown prospect of enhancing the performance of the traditional biometrics by fusing these new biometric modalities with established ones. Some of these biometric measures have shown tremendous potential in forensic applications. In this paper a review of two of these emerging biometric modalities is presented.

Texture based palm Print recognition using 2-D Gabor filter and sub space approaches

Palm Print based biometric systems have attracted much attention in various security applications because palm prints are rich with unique line, point and texture patterns which even low resolution palm scanners can easily capture. This paper presents a texture based palm print recognition method which employ 2D Gabor filter to extract texture information from the central part of hand and use subspace methods for dimension reduction. The test and training images are compared in terms of calculating Euclidean distance between them. Algorithm is tested on standard benchmark databases (CASIA and IIT Delhi) and the results clarify the effectiveness of our method in terms of the Correct Recognition Rate and Computation Time.

Knuckle Print Biometrics and Fusion Schemes -- Overview, Challenges, and Solutions

Numerous behavioral or physiological biometric traits, including iris, signature, hand geometry, speech, palm print, face, etc. have been used to discriminate individuals in a number of security applications over the last 30 years. Among these, hand-based biometric systems have come to the attention of researchers worldwide who utilize them for low- to medium-security applications such as financial transactions, access control, law enforcement, border control, computer security, time and attendance systems, dormitory meal plan access, etc. Several approaches for biometric recognition have been summarized in the literature. The survey in this article focuses on the interface between various hand modalities, summary of inner- and dorsal-knuckle print recognition, and fusion techniques. First, an overview of various feature extraction and classification approaches for knuckle print, a new entrant in the hand biometrics family with a higher user acceptance and invariance to emotions, is presented. Next, knuckle print fusion schemes with possible integration scenarios, and traditional capturing devices have been discussed. The economic relevance of various biometric traits, including knuckle print for commercial and forensic applications is debated. Finally, conclusions related to the scope of knuckle print as a biometric trait are drawn and some recommendations for the development of hand-based multimodal biometrics have been presented.

ECG based human authentication using synthetic ECG template

Ever increasing identity fraud in various online transactions has increased the need for the development of human authentication systems which are robust and provide complete population coverage. In this direction ECG based biometric systems have shown lot of promise as it inherently possesses the universality trait and does not require liveness detection unlike fingerprint or iris based systems. In this paper we present an approach which uses a synthetic ECG template for identity verification. The algorithm has been tested on in-house ECG database with a gap of two months between training and test samples. The proposed approach is computationally fast and simple.

Palmprint and Finger Knuckle Based Person Authentication with Random Forest via Kernel-2DPCA

This paper presents a hand biometric system by fusing information of palmprint and finger knuckle to check the loopholes that are present in transfer of payments through various levels of bureaucratic financial inclusion projects. Initially, a novel, fixed size ROIs of palm and finger knuckle has been extracted. The poor contrast ROI images are enhanced using modified CLAHE algorithm. To minimize the pose and illumination effects, Line Ordinal Pattern (LOP) based transformation scheme has been applied. The generation of dense feature representation by using dual tree complex wavelet transform can increase the discrimination power of independent local features. Then, the original feature space is mapped into high dimensional sub feature set, where K2DPCA is performed on each subset to extract high order statistics. Addressing to the matching problem, a high-performance Random Forest method has been employed. Finally, the two modalities are combined at weighted sum score level fusion rule which has shown the increased performance (CRR (100%), EER (0.68%), and (computation time (2130 ms)) of combined approach. The proposed method is evaluated using a virtual combination of publicly available PolyU palm print and PolyU FKP databases.

FKP based personal authentication using SIFT features extracted from PIP joint

Hand biometric systems have been adopted in a variety of access control application because hands are rich with unique line, point, vein, and texture patterns which even low resolution scanners can easily capture. In this article, finger dorsal surface is utilized to extract the local texture information by SIFT and further modified by LDA algorithm. Finally, the test and training images are compared in terms of employing a KNN classifier. The performance of proposed algorithm is tested over a standard benchmark FKP databases (PolyU) with different number of training samples. The obtained results clarify the effectiveness of proposed method with the other state-of-the-art algorithms in terms of the Correct Recognition Rate, Equal error rate, and Computation Time.

Comparison of PCA and 2D-PCA on Indian Faces

Face recognition is an extensively researched topic by researchers from diverse disciplines. Several unsupervised statistical feature extraction methods have been used in face recognition, out of these in this paper a comparison of the PCA(eigenfaces) and 2D-PCA approaches on Indian Faces has been presented. To test and compare their performances a series of experiments were performed on ORL database, Yale face database and then on an in-house dataset which has been collected over a span of 6 months. The performance parameters compared here are recognition rate and speed with varying number of training images. The application of various preprocessing techniques which can be used to improve their performance has also been studied.

Image hiding using unitary similarity transformation

A different and efficient method of image hiding is proposed here. It is based on unitary similarity transformation, involving calculation of eigen values and eigen vectors of a matrix and then transforming it into a diagonal matrix. Only secret image needs to be transformed and is then embedded to the cover image. Inverse transformation can be used to recover the secret image from the stego-image. The eigen vector matrix acts as decryption key. This algorithm is simple and can be easily implemented. It can greatly improve the security of the system, robustness of image-hiding. The quality of stego-image and the recovered image is improved up to a certain extent and is evident from the high PSNR of stego image.

Multimodal Biometric Authentication System Using Hand Shape, Palm Print, and Hand Geometry

Developing a multimodal biometric system based on single-shot imaging (SSI) has recently grown interested in researchers worldwide. A palm region basically enriches with most discriminative features like lines, shape, and geometry which can be easily clubbed and captured together. In this work, feature-level fusion of hand shape, geometry, and palm print features has been performed. The extracted palm ROI samples undergo certain rotation and illumination effects that limit the matching performance. ROI samples are first geometrically aligned and then transformed into illumination-invariant form using CS-LBP. Further, local key points of transformed ROI images are extracted using SURF descriptor. In addition to this, a set of novel geometrical and shape features have also been computed from the hand registered image. All the three set of features are concatenated, and then, the highly uncorrelated features are selected from the fused feature set using sub-pattern PCA for classification. The performance of the proposed multimodal system is found to be superior to each of individual modality as well as reported state-of-the-art systems.

Multimodal Biometric Authentication System Using Local Hand Features

In this work, the hand-based multimodal biometric system is presented using score-level fusion of hand geometry and local palmprint features. Initially, a palm ROI of fixed size has been cropped on the basis of finger base points. However, these images are not well aligned and reduce the matching accuracy. To better align them, L-K tracking-based palm image alignment method has been presented. Following this, the poor contrast ROI image is enhanced using novel fractional G-L filter. Then, local keypoints of aligned ROI images are extracted using Block–SIFT descriptor. Secondly, a set of novel geometrical features has been computed from Palmer region of hand image. Further, the highly uncorrelated features are selected from palm and hand geometry using Dia-FLD. In order to handle robust classification, a high-performance method Linear SVM has been used. Finally, score-level fusion rule has been employed which has shown the increased performance of combined approach in terms of Correct Recognition Rate (99.34%), Equal Error Rate (2.16%), and Computation Time (2084 ms). The proposed system has been tested on largest publicly available contact based and contactless databases: Bosphorus hand database, CASIA, and IITD palmprint databases to validate the results.