Research

Biometric Security

The technology that provides automated means for identifying people using their unique physiological body markers or behavioural characteristics.

Electrocardiogram (ECG) as a Biometrics

Biometric authentication systems are being used with a trust to support identification experts in the digital society. But securing the system from fraudulent attacks is a great challenge. The security of a system can be threatened by presenting a non-live or fake biometrics sample for authentication. Although, different techniques have been proposed in the past those address the problem of vitality detection from biometrics but these techniques are far away from the final solution. Our research explore the feasibility of the physiological signals to use as biometrics. In particular, our focus is on the Electrocardiogram (ECG) as such its potential to use as a candidate of biometrics especially vitality-enabled biometrics. We developed a vitality enabled novel identity recognition system using the ECG signal. We can easily visualize the subject dependent features in the ECG signal collected in different sessions.

Multimodal Biometrics

The biometric systems rely on the evidence of any single biometric information may not achieve the desired performance due to noisy data, non-universality, lack of uniqueness of the chosen biometric trait. Most importantly, they are not robust enough against fraudulent attacks. The performance of the a biometric system can be improved by combining the multiple biometric characteristics and to use them for identity verification, simultaneously. These systems are referred as multimodal or multibiometric systems. We have developed a novel multibiometric system that fuses the ECG with face and fingerprint biometrics. The advantage of aforementioned system is that it would achieve the optimum performance. In addition, it may provide a strong protection against falsification to be enrolled in the recognition system because ECG has an inherited feature of vitality that signifies the life signs.

Biometrics Template Protection

Biometrics is used for secured authentication of individuals but the system itself needs protection from spoof attacks. It stores the characteristics of an individual's biometrics in the form of templates. These templates are the targets for the adversary to get holds and leak them. The templates may also contain personal information such as date of birth, contact detail along with age, gender and ethnicity of individuals. Therefore, people are more concerned to security and privacy of their personal information while sharing their biometrics data to any third party. Therefore, template protection is not only needed for biometric system preservation, but also for security and privacy against theft and attacks.

In this research we are exploring the traditional techniques of biometric template protection such as biometric cryptosystems and template transformations along with other state-of-the-art techniques using deep learning.

Social Media Analytics

Social media are Internet based applications that are used for sharing information of social, economic and political interests among virtual communities. The prominent resources of social media include Facebook, LinkedIn, Twitter, Instagram, WeChat, WhatsApp, YouTube, Google+, Snapchat and Pinterest etc. The social media provides a range of benefits and opportunities to empower people and communities in a variety of ways. The social media has become a crucial handler for acquisition and dissemination of information across various domains such as science, entertainment, business, politics, academics, healthcare and crisis management. The social media has became an integral part of digital society with the proliferation of social websites and applications thus, resulting in explosion of data during the recent years.

Social media analytics is the technique of acquiring and analyzing data from social networks. The aim of this research is to device intelligent techniques for social media analytics that may address the challenges of different applications of social media. The focus of this research to understand and analyse the social media data by computation model using AI/ML techniques. The developed frameworks for social media intelligence platforms can provide aid to public safety by gathering information from social networks. Further, the system may predict potential activities that could help in preventing the antisocial activities for maintain public harmony.

Medical Image Analysis & Classification of Diseases

Medical imaging is the technique to study and analyse images of different body parts and organs for medical uses in particular to identify and/or study diseases. The principal modalities use in medical image analysis are ECG, EEG, X-rays, USG, CT, MRI, f-MRI and PET etc. The medical image analysis seeks to reveal internal structures hidden by the skin and bones, as well as to diagnose and treat diseases.

The advancement of computing technologies especially, artificial intelligence, machine learning, pattern recognition. computer vision, robotics and deep learning provide an umbrella to researchers to automate the process of clinical diagnosis of diseases as well as to monitor their progress. The computer aided diagnosis and classification of diseases are posing bigger challenges for open research. The aim of this research to develop intelligent systems that automatically identify the disease patterns from medical images and efficiently classify them for better clinical care.

Healthcare & Technology - Smart Healthcare

Healthcare is one of the most dynamic area that faces major challenges of quality, accessibility and affordability for a large section of the population. Modern digital age using the applications of artificial intelligence (AI) and machine learning (ML) influences our daily life. The intelligent computing system is being used with the medical industry with no exception. The objective of this research is to take advantages of using Information and Communication Technology (ICT) and computing technology to improve the quality of healthcare, turning traditional into smart healthcare.

The smart healthcare consists of the use of technology for good care of patients, utilizing better diagnostic tools, availability of doctors, use of smart devices to collect patient specific data for better quality of life to them anytime and anywhere. Therefore the use of computing model with intelligent analysis of medical records to facilitate health services to needy people is the aim of this research. It is the convergence of information technology and medical informatics with computer. The interesting outcomes like Internet of Medical Things (IoMT) when AI combines with IoT the new nervous system for healthcare would be future direction of research.

Computer Vision

The area of computer vision provides the window to computers to understand and analyse the visual world. Using the techniques of AI, ML, PR and deep learning the machine can able to identify and classify objects more accurately than a normal human being.

In computer vision we are interested in object recognition, event detection and actions classification in moving objects. The goal of this work is to efficiently detect the object categories from moving scenes. While performing actions the emphasis of this research is to identify and classify human actions available in the scene.