Skip to main content


Identification Using Biometric Traits

Different biometrics traits as well as soft biometrics can be used to identify individual. Examples of biometric traits include fingerprint, face, lips, iris, palm-print, vein, dental, retina, hand geometry, tattoo, soft biometrics, voice, ear, signature, DNA, gender, age, ethnicity, keystroke and gait.

Example of three typical projects that have been done in BIIC are described below: At BIIC we have collected and investigated different biometric traits with considerable success to identify individuals.

Face Recognition with Significant Aging and Photo Quality Variations

Human identity management is important in visual surveillance, security, and law enforcement applications. As an important biometric cue, face recognition is very useful for non-invasive identification of non-cooperative users. However, there are great challenges in developing robust face recognition systems.

The great challenges include facial aging variations and image quality changes among others. For example, in the enrollment, a high quality face photo can be captured, while in recognition, the query face images can be captured by cameras with different qualities (i.e., inter-platform operation), or appeared in different formats, e.g., passports, social media, photo scan, newspapers, etc. Furthermore, the subjects in query can often have ages different from the gallery with significant facial appearance changes. Thus in many real applications, face recognition suffers from coupled aging variations and all kinds of photo quality changes. In this research project, we have developed new methods for face matching robust to aging and photo quality variations. 

In this project, several tasks were explored:

  1. We developed a quantitative measure of image quality based on spatial and frequency domain analysis, and built the relation between the matching performance and the quality measure.
  2. Developed a robust face matching method based on correlation analysis. Humans are good at recognizing face images with low-quality and aging variations, although it is extremely hard for a computer. There must be some common attributes among the face images of the same subjects in human perception. Based on this, we developed a “correlation” measure to characterize the common attributes among faces of the same subjects.
  3. A multi-view statistical learning algorithm for face matching was also developed. A novel perspective to deal with aging and quality variations is to model them as a multi-view recognition problem, where each face image is one particular “view” of the subject. Therefore, we developed a multi-view learning method to cope with the aging and photo quality variations.
  4. We evaluated the proposed approaches on the FBI BCOE database, which was collected recently at WVU with 1,099 subjects having significant aging and various photo quality changes. This database is much larger than the FG-NET database (82 subjects). It has not been used before for computational face recognition.

Fusing Biometric and Biographic Information in Identification Systems

Multi-biometric systems consolidate evidence provided by multiple sources to establish the identity of an individual. Typically, these sources correspond to the biometric traits of an individual such as face, iris, fingerprint and palmprint. In this project we investigated the problem of combining non-biometric, biographic data (such as name, age, gender, ethnicity, nationality, etc.) with biometric information in order to render better decisions about the identity of individuals. Such a research effort is warranted due to the potential use of mixed data (i.e., biometric and non-biometric) in large-scale identity management systems such as US-VISIT (now OBIM), TWIC and E-VERIFY. Further, social media sites such as facebook, linked-in, and twitter have both biometric (viz., face images) and biographic details of an individual.  However, the role of biographic data in establishing the identity of an individual in large-scale systems has hitherto not been studied.

The main objective of this project was to undertake a systematic study to establish the utility of combining biographic data with biometric information to establish identity. 

The four tasks in this project were to:

  1. Perform a statistical analysis to understand the contribution of biographic data such as name, gender, nationality and ethnicity in establishing the identity of an individual
  2. Understand the distribution and variation of names across nationalities
  3. Design a mechanism to retrieve the identity of an individual from a database based on input biographic data (i.e., query). The mechanism will accommodate data (in the database and/or in the query) that is noisy (e.g., misspelt name) or incomplete (e.g., missing biographic information). Design a fusion mechanism that combines biographic data with biometric data in order to establish the identity of an individual. The fusion mechanism will account for “noise” in both types of data (e.g., incorrect or missing data).

Incorporating Biological Models in Iris Anti-Spoofing Schemes

For over a decade, spoof detection (i.e., anti-spoofing) has been an important topic within the biometrics community. In the area of iris anti-spoofing, researchers have focused their attention on either the static properties of the iris tissue (e.g., for detecting contact lenses and printed spoofs) or the dynamic aspects of the iris (e.g., pupil dilation). However, a comprehensive evaluation is unavailable since individual approaches address specific factors while ignoring others.

This project aims to link these different approaches – empirically and biologically – to provide a comprehensive understanding while engineering universal security processes that mitigate vulnerabilities to spoofing. The results of this research can help advance iris recognition technology by improving current iris recognition algorithms as well as providing further insight into how anti-spoofing evaluations must be conducted on this topic.

Three main tasks of this project are:

  1. evaluate the correlation between static and dynamic ocular properties based on empirical evidence. The WVU PLR database will be used for this study.
  2. Combine both static and dynamic perspectives, including biological modeling, for designing robust iris anti-spoofing schemes.
  3. Address the following main questions:
    1. What are the limitations of the proposed approaches when used independently and collectively?
    2. How well are they at characterizing the effects of a live iris and a spoof?
    3. What are additional aspects of spoof vulnerability that are not incorporated in the proposed model?