Biometric-based Identification Research Articles

Utilizing EfficientNet for sheep breed identification in low-resolution images

Automatically recognizing sheep breeds is highly valuable for the sheep farming industry, allowing farmers to pinpoint their specific business needs. Accurately distinguishing between sheep breeds poses a challenge for numerous farmers with limited expertise. Although biometric-based identification offers a feasible solution, its application becomes impractical when assessing large numbers of sheep in real-time. Therefore, the implementation of an automatic sheep classification model that can replicate the breed identification skills of a sheep breed expert can come in handy. This would be particularly beneficial for novice farmers who could utilize handheld devices for breed classification. To address this objective, we propose employing a convolutional neural network (CNN) model capable of rapidly and accurately identifying sheep breeds from low-resolution images. Our experiment utilized a dataset of 1680 facial images representing four distinct sheep breeds. We conducted experiments on the dataset using various EfficientNet models and found that EfficientNetB5 achieved the highest performance with 97.62 % accuracy on a 10 % test split. The classification model we developed has the potential to assist sheep farmers in efficiently distinguishing between different breeds, facilitating more precise assessments and sector-specific classification for various businesses within the industry.

Feature Extraction of the Human Ear Based on Enhanced Active Contour Method

Human ear detection is one of the key features in biometric-based identification and person authentication. Detecting ears for position, dimension, and texture is tedious due to the height and movement of humans. The biometric systems based on ear detection are defaced by the height of humans and specific positions. For this purpose, a multi-feature and multi-contour-based assessment for precise detection is required. To address this problem, this article introduces an Enhanced Active Contour Method (EACM) for improving ear detection precision. The proposed method extracts the textural features for differentiating the antitragus and helix of the human ear. The extracted features are fused using a fuzzy threshold for the maximum possibilities of the ear structures. In verifying the maximum possibilities, the helix and antitragus contours are fused by identifying similar features. The features from the existing dataset are used for verifying the similarity; the similarity succeeding contours are used for detecting human ears. The proposed method maximizes accuracy and precision by 12.99% and 12.79% and reduces extraction time by 11.34%.

Automatic identification of individual yaks in in-the-wild images using part-based convolutional networks with self-supervised learning

Yaks (Bos grunniens) are the most important domestic animals for people living at high altitudes. In order to implement precise livestock management for yaks, it is of significant importance to automatically identify, keep track of, and monitor yaks. Traditional animal identification methods such as ear tags, tattoos, and RFID based methods suffer from problems like animal infection, high maintenance cost, inefficiency or sensor failure. Existing biometric-based identification methods for livestock such as muzzle prints, iris patterns, and retinal vascular patterns mostly require that animals are under control, either technically or physically, and are thus costly to deploy especially for yaks which are loosely raised on the grassland pastures and migrate with the seasons. In this paper, we propose a novel method for identifying individual yaks in in-the-wild images captured under unconstrained conditions. We utilize the part-based convolutional network (PCN) to obtain discriminative part-level feature representations. To further enhance the feature discriminativeness and alleviate the impact of small amount of yak image data, we implement self-supervised learning strategy by proposing random erasure and region-visibility prediction (RERP) as an auxiliary learning task. Experiments performed on the YakReID-103 dataset demonstrate that (i) when left and right side views of yaks are treated separately, the Rank-1 accuracy and mAP achieved by the proposed method with SEResNet50 backbone are up to 97.57% and 76.30%, which significantly advance the state-of-the-art, and (ii) when generalizing to different views, the proposed method with ViT backbone again obtains the best results compared with the counterpart methods.

A Unified Local-Global Feature Extraction Network for Human Gait Recognition Using Smartphone Sensors.

Smartphone-based gait recognition has been considered a unique and promising technique for biometric-based identification. It is integrated with multiple sensors to collect inertial data while a person walks. However, captured data may be affected by several covariate factors due to variations of gait sequences such as holding loads, wearing types, shoe types, etc. Recent gait recognition approaches either work on global or local features, causing failure to handle these covariate-based features. To address these issues, a novel weighted multi-scale CNN (WMsCNN) architecture is designed to extract local to global features for boosting recognition accuracy. Specifically, a weight update sub-network (Ws) is proposed to increase or reduce the weights of features concerning their contribution to the final classification task. Thus, the sensitivity of these features toward the covariate factors decreases using the weight updated technique. Later, these features are fed to a fusion module used to produce global features for the overall classification. Extensive experiments have been conducted on four different benchmark datasets, and the demonstrated results of the proposed model are superior to other state-of-the-art deep learning approaches.

Seeing Like an Infrastructure: Low-resolution Citizens and the Aadhaar Identification Project

Large-scale data systems play a central role in how state bureaucracies come to know and manage citizens. Such systems are endemically uneven in implementation, producing distributed and exclusionary consequences that are among their most important effects. These systems also operate as infrastructures in the rich and complex sense of the term that has been a core contribution of CSCW to the wider computing and social science fields. Building on James Scott's work on 'seeing like a state', we conceptualize 'seeing like an infrastructure' as a more supple analytic perspective that maps the distributed work and uneven consequences through which designers, bureaucrats, and users (here, citizens) assign or claim representation in the consequential data systems that increasingly shape and define citizenship. Drawing on eighteen months of ethnographic fieldwork into Aadhaar, India's biometrics-based identification project, and studies of infrastructure, marginalization, and citizenship in CSCW and allied fields, we argue that this perspective provides crucial insight into the strategies and mechanisms by which effective access to the basic rights and entitlements of citizenship are granted, claimed, and at times undermined. More specifically, we show how challenges in implementing Aadhaar's three key processes-enrollment, seeding, and authentication-give rise to a spectrum of resolution in which the rights and entitlements of 'high-resolution citizens' are expanded, while those of 'low-resolution citizens' are curtailed.?

ActID: An efficient framework for activity sensor based user identification

Identification is the core of any authentication protocol design as the purpose of the authentication is to verify the user’s identity. The efficient establishment and verification of identity remain a big challenge. Recently, biometrics-based identification algorithms gained popularity as a means of identifying individuals using their unique biological characteristics. In this paper, we propose a novel and efficient identification framework, ActID, which can identify a user based on his/her hand motion while walking. ActID not only selects a set of high-quality features based on Optimal Feature Evaluation and Selection and Correlation-based Feature Selection algorithms but also includes a novel sliding window based voting classifier. Therefore, it achieves several important design goals for gait authentication based on resource-constrained devices, including lightweight and real-time classification, high identification accuracy, a minimum number of sensors, and a minimum amount of data collected. Performance evaluation shows that ActID is cost-effective and easily deployable, satisfies real-time requirements, and achieves a high identification accuracy of 100%.

Hand Side Recognition and Authentication System based on Deep Convolutional Neural Networks

The human hand has been considered a promising component for biometric-based identification and authentication systems for many decades. In this paper, hand side recognition framework is proposed based on deep learning and biometric authentication using the hashing method. The proposed approach performs in three phases: (a) hand image segmentation and enhancement by morphological filtering, automatic thresholding, and active contour deformation, (b) hand side recognition based on deep Convolutional Neural Networks (CNN), and (c) biometric authentication based on the hashing method. The proposed framework is evaluated using a very large hand dataset, which consists of 11076 hand images, including left/ right and dorsal/ palm hand images for 190 persons. Finally, the experimental results show the efficiency of the proposed framework in both dorsal-palm and left-right recognition with an average accuracy of 96.24 and 98.26, respectively, using a completely automated computer program.

An investigation of the predictability of the Brazilian three-modal hand-based behavioural biometric: a feature selection and feature-fusion approach

New security systems, methods or techniques need to have their performance evaluated in conditions that closely resemble a real-life situation. The effectiveness with which individual identity can be predicted in different scenarios can benefit from seeking a broad base of identity evidence. Many approaches to the implementation of biometric-based identification systems are possible, and different configurations are likely to generate significantly different operational characteristics. The choice of implementational structure is, therefore, very dependent on the performance criteria, which is most important in any particular task scenario. The issue of improving performance can be addressed in many ways, but system configurations based on integrating different information sources are widely adopted in order to achieve this. Thus, understanding how each data information can influence performance is very important. The use of similar modalities may imply that we can use the same features. However, there is no indication that very similar (such as keyboard and touch keystroke dynamics, for example) basic biometrics will perform well using the same set of features. In this paper, we will evaluate the merits of using a three-modal hand-based biometric database for user prediction focusing on feature selection as the main investigation point. To the best of our knowledge, this is the first thought-out analysis of a database with three modalities that were collected from the same users, containing keyboard keystroke, touch keystroke and handwritten signature. First, we will investigate how the keystroke modalities perform, and then, we will add the signature in order to understand if there is any improvement in the results. We have used a wide range of techniques for feature selection that includes filters and wrappers (genetic algorithms), and we have validated our findings using a clustering technique.

Face geometry as a biometric-based identification system

In this paper, we present a method for identifying a person with a biometric-based on face geometry. This person identification system design has been done using face biometrics by implementing feature extraction of distances between facial features. Face biometrics were chosen because they have unique characteristics and do not change in each person. In this method, a system analysis that separates the face image into face components, including the eyes, nose, and mouth on the face image is taken from the front view position. Each component of the feature that has been detected is measured by its distance to form the face semantics. The steps taken in image processing begin with determining the region of interest (ROI). Furthermore, the results of cropping ROI feature extraction are done to get the eye, nose and mouth. After that, a centroid is determined for each feature so that the distance between features can be calculated using the equation of the distance between the two coordinates, so we get 8 distances as matching parameters. The matching process is done using euclidean distance, which is calculating the smallest distance between the test image and the database. Based on research results, this system can be used to identify someone using face biometrics with fixed poses and expressions. The accuracy of this system is 100 percent

Give Me a Database and I Will Raise the Nation-State

This paper draws on an ethnographic study of Aadhaar, India’s biometrics-based national identification infrastructure, to investigate how members of its design team conceptualised and understood their techno-bureaucratic enterprise of assigning unique numbers to Indian residents. Members described their work using the metaphor of building an hour-glass: the Aadhaar number and its authentication services constitute the waist of this hour-glass; below were innovations in biometric devices and above were applications requiring identity verification services. They believed the entire ecosystem sustaining Aadhaar could be controlled by prescribing interactions between the waist and the components above and below. They extended this metaphor to reimagine the Indian government as a platform of services controlling only a specific part of a service—the waist—while opening space for innovation by integrating it with other market services above and below. This paper documents the emergence of this imaginary of ‘platformised’ government collecting real-time citizen data to support personalised state–citizen interactions to unpack how the future(s) of Indian government services shapes and is shaped by it. Such future(s) constitute the Indian population as a database, bureaucracies as centralised dashboards, and government as arbiter in the circulation of citizen data.

Research of Face Recognition with Fisher Linear Discriminant

Face identification systems are developing rapidly, and these developments drive the advancement of biometric-based identification systems that have high accuracy. However, to develop a good face recognition system and to have high accuracy is something that’s hard to find. Human faces have diverse expressions and attribute changes such as eyeglasses, mustache, beard and others. Fisher Linear Discriminant (FLD) is a class-specific method that distinguishes facial image images into classes and also creates distance between classes and intra classes so as to produce better classification.

Using biometric-based identification systems in Brazil: A review on low cost fingerprint techniques on-the-go

Automatic authentication has become an essential service in several public areas. However, although the technology related with this kind of service has evolved, the price tag of its use is not affordable for most countries. In the so-called “under developed” counties, such as Brazil, South Africa and India, for example, registration systems are often paper-based and/or cover only a fraction of the population. Thus, the reality is that there is an increasing gap into the usage of such technologies amongst different countries and it can be a factor that makes development more difficult and, therefore, less inclusive. One of the main technologies used for automatic identity prediction is based on biometrics analysis, which can distinguish physical or behavioural features to help overcome the traditional paper-based identity systems. Despite the limitations already mentioned, Brazil is known to have introduced several different uses of biometric-based technologies for authentication. However, the use of these technologies is not always ideal and, since the population size is a key factor, it is essential to select the most affordable option which is not necessarily the most adequate for the country's needs. This paper will focus on establishing what biometric-based solutions exist in Brazil today, highlighting the main challenges, as well as briefly proposing a new prototype for mobile fingerprint acquisition.

Gabor Filter and Texture based Features for Palmprint Recognition

In this paper, we propose an efficient personal identification system based on palmprint recognition. Palmprint is widely used in biometric-based identification system. Palmprint is robust and obtained in a simple way. After extracting region of interest (ROI), the ROI is passed through Gabor filters with different wavelengths and orientations. Then, binarized statistical image features (BSIF) of phase of outputs of Gabor filters are obtained. Different BSIF codes are combined together and then, the histogram of final BSIF code is calculated. Efficient features from histogram are calculated and are given to the K-nearest neighbor (KNN) classifier to perform personal identification. Experimental results on PolyU database demonstrate that proposed algorithm achieves the higher accuracy than the recently proposed algorithms.

Multimodal biometric-information fusion using the Radon transform

This paper proposes multimodal biometric-based personal identification. Palmprint and fingerprint modalities are utilized in the proposed model due to the availability of more distinctive features and the ease of capture. Principle lines are the main discriminative features in the palmprint, whereas orientations of the ridge and valley structures are the main features by which to identify the fingerprint. To extract these features, the use of the Radon transform is proposed in this work. However, the Radon transform is sensitive to the orientation. In order to make the model rotation invariant and insensitive to noise, a normalization process is generally used. Here, a logarithm-based normalization process has been utilized in the proposed model. A Euclidean-based matching process that is invariant to the rotation has been used. The proposed model is applicable to low resolution images and is invariant to rotation, insensitive to noise, and has less computational complexity as compared to other models.

English

With the fast development of video hardware and software in recent years, intelligent video systems have been widely used in industry, transportation, security, etc. at the same time, a lot of biometric technologies, comprising automated methods for uniquely recognizing people based on their physical or behavioural traits, such as face, fingerprint, palm print, finger-knuckle-print, gait, are also based on video or image analysis. Iris recognition is emerging as one of the important methods of biometrics-based identification systems. Several crucial factors Gait Keystroke pattern Signature of iris biometrics include rich and unique textures, non-invasiveness, and stability of iris pattern throughout the person’s lifetime, public acceptance, and availability of user friendly capturing devices. These factors have attracted the researchers to work in this evolving field over the past decade. The iris recognition consists of iris localization, normalization, encoding and comparison. In this paper, segmentation of periocular features and the encoding part of iris recognition is analysed.

Learning Race from Face: A Survey.

Faces convey a wealth of social signals, including race, expression, identity, age and gender, all of which have attracted increasing attention from multi-disciplinary research, such as psychology, neuroscience, computer science, to name a few. Gleaned from recent advances in computer vision, computer graphics, and machine learning, computational intelligence based racial face analysis has been particularly popular due to its significant potential and broader impacts in extensive real-world applications, such as security and defense, surveillance, human computer interface (HCI), biometric-based identification, among others. These studies raise an important question: How implicit, non-declarative racial category can be conceptually modeled and quantitatively inferred from the face? Nevertheless, race classification is challenging due to its ambiguity and complexity depending on context and criteria. To address this challenge, recently, significant efforts have been reported toward race detection and categorization in the community. This survey provides a comprehensive and critical review of the state-of-the-art advances in face-race perception, principles, algorithms, and applications. We first discuss race perception problem formulation and motivation, while highlighting the conceptual potentials of racial face processing. Next, taxonomy of feature representational models, algorithms, performance and racial databases are presented with systematic discussions within the unified learning scenario. Finally, in order to stimulate future research in this field, we also highlight the major opportunities and challenges, as well as potentially important cross-cutting themes and research directions for the issue of learning race from face.

Dorsal hand vein recognition based on directional filter bank

With the increasing needs of security systems, vein recognition is reliable as one of the important solutions for biometrics-based identification systems. The obvious and stable line-feature-based approach can be used to clearly describe dorsal hand vein patterns for personal identification. In this paper, a directional filter bank involving different orientations is designed to extract vein patterns and the minimum directional code is employed to encode line-based vein features into binary code. In addition, there are many non-vein areas in the vein image, which are not meaningful for vein recognition. To improve accuracy, the non-vein areas are detected by evaluating the variance of the minimum directional filtering response image and are considered as non-orientation code. In total, 4,280 dorsal hand vein images from 214 persons are used to validate the proposed dorsal hand vein recognition approach. A high accuracy (\(>\)99 %) and low equal error rate (0.54 %) were obtained using the proposed approach, which shows that the approach is feasible and effective for dorsal hand vein recognition.

Gaussian directional pattern for dorsal hand vein recognition

With the increasing needs in security systems, hand vein recognition is reliable as one of the important solutions for biometrics-based identification systems. This work presents an effective approach for dorsal hand vein recognition by analysing vein patterns. The methodology involves extraction of dorsal hand vein features based on Gaussian filter. In order to obtain effective pattern of dorsal hand vein vascular, we propose an innovative and robust Gaussian directional filter method to extract the dorsal hand vein patterns and to encode the vein features in binary code. A new coding scheme, namely Gaussian directional binary code, is then proposed for dorsal hand vein recognition. To evaluate the efficacy of the proposed approach, the normalised Hamming distance used in recognition is adopted. A total of 5120 dorsal hand vein images were collected from 256 persons to verify the validity of the proposed dorsal hand vein recognition approach. High accuracies (>99%) and low equal error rate (0·95%) have been obtained by the proposed method. Experimental results demonstrate that our proposed approach is feasible and effective for dorsal hand vein recognition.

A secure palm vein recognition system

With the increasing needs in security systems, vein recognition is one of the important and reliable solutions of identity security for biometrics-based identification systems. The obvious and stable line-feature-based approach can be used to clearly describe a palm vein patterns for personal identification. In this paper, a directional filter bank involving different orientations is designed to extract the vein pattern and the minimum directional code (MDC) is employed to encode the line-based vein features in binary code. In addition, there are many non vein pixels in the vein image and those pixels are unmeaning for vein recognition. To improve the accuracy, the non-vein pixels are detected by evaluating the directional filtering magnitude (DFM) and considered the non-orientation code. A total of 5120 palm vein images from 256 persons are used to verify the validity of the proposed palm vein recognition approach. High accuracies (>99%) and low equal error rate (0.54%) obtained by the proposed method show that our proposed approach is feasible and effective for palm vein recognition.

Large Visual Repository Search with Hash Collision Design Optimization

Visual search over large image repositories in real time is one of the key challenges for applications such as mobile visual query-by-capture, augmented reality, and biometrics-based identification. Search accuracy and response speed are two important performance factors. This article focuses on one of the important elements of this technology that enables large-scale visual search: indexing (or hashing). Indexing is the process of organizing a database of searchable elements into an efficiently searchable configuration. The searchable elements in our case are compact features extracted from images. This article explores a new indexing scheme. The authors optimize the design of a hash-code collision and counting scheme to enable fast search of visual features of MPEG CDVS.