Skip to main content

Ongoing Research

Detecting Illicit Drug Abuse for Improved Face Recognition

Illicit drug abuse has become one of the primary health and social concerns in today’s world. The problem of illicit drug abuse is becoming more apparent, considering that 1 out of 10 drug users is a problem drug user, who is suffering from drug dependence. Illicit drug abuse and its detection has applications in health and medical areas as well as in face recognition. In this research, we demonstrate the impact of facial variations caused due to illicit drug abuse on face recognition. We also propose a non-invasive classification algorithm using dictionary learning to detect face images affected due to illicit drug abuse such that it can be used in conjunction with current face recognition systems.

Relevant Publications

Recognizing Faces Altered by Plastic Surgery

Researchers have shown that the changes in face features due to plastic surgery can be modeled as a covariate that reduces the ability of algorithms to recognize a person's identity. We propose a novel solution to discriminate plastic surgery faces from regular faces by learning representations of local and global plastic surgery faces using multiple projective dictionaries and by using compact binary face descriptors. 

Relevant Publications

Kinship Verification using Faces

Kinship verification has a number of applications such as organizing large collections of images and recognizing resemblances among humans. A hierarchical Kinship Verification via Representation Learning (KVRL) framework is utilized to learn the representation of different face regions in an unsupervised manner. We propose a novel approach for feature representation termed as filtered contractive deep belief networks (fcDBN). The proposed feature representation encodes relational information present in images using filters and contractive regularization penalty. A compact representation of facial images of kin is extracted as an output from the learned model and a multi-layer neural network is utilized to verify the kin accurately.

Relevant Publications

Detecting Medley of Iris Spoofing Attacks

Iris is one of the most reliable and accurate biometric modalities due to the highly unique character of iris tissue structure. It has been used successfully in numerous applications including national ID projects and border security. The success of large scale identity applications using iris recognition, in turn, means there are now individuals who can gain advantage by defeating these applications to gain unauthorized access to locations or resources or to escape recognition as an individual of interest. In real-world scenarios, iris recognition systems have to handle and detect various types of spoofing attacks including fake/printed iris images, synthetic images, and textured contact lenses. This research presents a novel framework utilizing structural and textural features to detect multiple complex spoofing attacks.

Relevant Publications

Biometrics based CAPTCHA for Improved Security

CAPTCHAs (Completely Automated Public Turing Test to tell Computers and Humans Apart) have been a common tool for preventing unauthorized access to websites for over a decade. However, increasingly sophisticated optical character recognition algorithms and attack strategies have rendered traditional CAPTCHAs insecure. In this research, we have designed a new CAPTCHA incorporating biometric information such as face detection and face recognition. The proposed CAPTCHA is language independent and easy to solve by humans but difficult to attack by automated algorithms.

Relevant Publications: